Multitrack Linearly Polarized Spectrometer for Simultaneous Kinetic UV–Vis, Polarization-Resolved- Scattering, and Photoluminescence Measurements

Abstract

Dynamic systems, defined by their continuous temporal evolution, are central to advancements in chemistry, biology, and materials science. Optical techniques that leverage light absorption, scattering, and emission are essential for characterizing structural and property changes in these systems. However, conventional optical toolssuch as UV–vis spectroscopy, fluorescence, and scattering techniquesprovide fragmented or incomplete insights, making it challenging to comprehensively understand dynamic processes and ensure reliable data interpretation. Herein, we introduce a charge-coupled device (CCD)-based multitrack linearly polarized spectrometer (MLPS) designed for simultaneous kinetic UV–vis, polarization-resolved scattering, and photoluminescence measurements. The MLPS facilitates concurrent quantification of scattering and fluorescence intensities and depolarizations, alongside UV–vis extinction, with subsecond temporal resolution. By integrating high temporal resolution with the ability to capture complementary spectra, the MLPS significantly enhances the functionality of optical spectroscopy, paving the way for broader applications in dynamic system analysis and advancing research across multiple scientific disciplines. Furthermore, the instrument characterization and data preprocessing methodologies presented here provide valuable insights for the future development of multitrack CCD-based spectrometers.

1. Introduction

Optical spectroscopy serves as a fundamental technique in chemical, biological, and materials research, providing crucial insights into material properties and their dynamic behaviors. − By exploiting light–matter interactions, including absorption, scattering, and emission, spectroscopic techniques such as UV–visible (UV–vis) spectrophotometry, dynamic and static light scattering, and spectrofluorometry have become indispensable tools across numerous scientific fields, including nanotechnology, materials science, environmental science, and biomedical diagnostics. − These methods are highly valued due to their nondestructive nature, coupled with high spectral and temporal resolution, and versatility across diverse experimental settings, making them essential for both fundamental research and application studies.

Collectively, these optical techniques provide substantial information on the photophysical properties and morphological characteristics of materials. UV–vis absorption spectroscopy is commonly used to determine optical bandgaps in semiconductors and elucidate electronic states in organic dyes. − Dynamic light scattering (DLS), as a scattering-based technique, is widely used to determine particle size distributions, − since scattering intensity and depolarization are highly sensitive to a material’s size, shape, and aggregation state, making them essential for characterizing nanoparticles, colloids, and complex fluids, while polarized light scattering offers valuable insights on shape anisotropy and molecular alignment in structured materials. − Fluorescence spectroscopy is widely used for investigating quantum dots, fluorescent proteins, and molecular probes in imaging and sensing applications. − Meanwhile, fluorescence depolarization, also known as fluorescence anisotropy, provides insights into fluorophore rotational mobility (rotational diffusion rate) and excited-state lifetimes, information directly related to the molecular orientation, aggregation behavior, and environmental interactions. − While individual spectroscopic techniques provide valuable information, their inherent limitations necessitate the integration of multiple optical methods for a more comprehensive characterization of materials. − These approaches are effective for steady-state systems where sequential measurements can be applied to the same system. However, it poses challenges when addressing dynamic systems where the materials’ chemical, morphological, and photophysical properties vary over time.

The information deficiency with existing tools also compromises the reliability of the data interpretation. One prominent example is UV–vis spectrophotometry. While this popular tool measures sample UV–vis extinction, it offers no insights into the signal origins, absorption, scattering, or both of the latter. − A further complication is the cascading photomatter interactions (CPMIs), the sequential photon–matter interactions triggered by the same incident photon, that are ubiquitous in practical optical spectroscopic applications. − With the only exception of a small number of solutions that can be approximated as pure absorbers, CPMI is inherent in other types of solutions including the ones that can be approximated as pure scatterers, simultaneous absorbers and scatterers, and simultaneous absorbers and emitters and those that simultaneous absorb, scatter, and emit light. Example CPMIs include the absorption or rescattering of the scattered light and the secondary fluorescence emission triggered by absorption of primary fluorescence photons generated by the absorption of the incident light. Such CPMIs drastically complicate the analysis of the experimental data obtained with conventional optical tools, − often leading to ambiguous data interpretations.

To overcome these limitations, we developed a charge-coupled device (CCD)-based multitrack linearly polarized spectrometer (MLPS) that employs free-space excitation and fiber-coupled spectral acquisition. By design, this system enables simultaneous acquisition of four spectra: a reference spectrum for monitoring the excitation intensity variation; a forward spectrum that detects the excitation light passing through the sample solutions; and two sideward linearly polarized spectra differing in their detection polarization directions. This integrated system enables concurrent quantification of the UV–vis extinction, polarization-resolved scattering, and fluorescence intensities and depolarizations, all with the same sample solutions with a subsecond temporal resolution.

Herein, we focus on the MLPS instrument design and characterization, data preprocessing and analysis protocols, and a proof-of-concept demonstration of the MLPS characterization of porphyrin self-assembly, a dynamic system used for performance evaluation. CCD is widely utilized in optical spectroscopy due to its high sensitivity, broad spectral range and the ability to capture parallel spectral data efficiently. To date, the CCD has been used for multitrack UV–vis, − fluorescence, − and polarization-resolved spectroscopic analysis. − However, to the best of our knowledge, MLPS is the first spectroscopic tool enabling the concurrence of UV–vis and polarization-resolved light scattering and fluorescence measurements. While the CCD greatly expanded optical spectroscopy capabilities, it has multiple challenges that can complicate the data analysis. Such complications include cosmic spike interference, saturation limits due to CCD pixel well depth, and spectral distortion due to differences in pixel responsivities. Additional challenges for multitrack spectroscopic analysis include spectral cross-talk and wavelength mismatches among the data obtained with different spectral tracks.

Commercial CCD-based spectrometers generally include built-in methods for cosmic spike removal. Even though these methods work effectively for single-pixel cosmic events, they fail when the cosmic spikes are so intense that they bleed to several neighboring pixels. The upper-bound-spectrum method is highly effective in identifying and removing cosmic spikes with no assumption of the pixel width of the cosmic spikes, but it requires multiple spectral acquisitions. − Therefore, this method is applicable only to steady-state samples where the sample spectral features remain constant during the sequential spectral acquisitions but not to dynamic systems. To address this issue, a new time-domain cosmic spike removal algorithm is developed herein in kinetic spectral acquisition (vide infra).

The ultimate performance objective of the MLPS instrument is to simultaneously capture spectral information equivalent to that provided by combined UV–vis, linearly polarized resonance synchronous (LPRS) spectroscopy, and linearly polarized anti-Stokes-shifted, on-resonance, and Stokes-shifted (LPAOS) spectroscopy. LPRS and LPAOS are two recent linearly polarized spectroscopic methods developed with a commercial spectrofluorometer equipped with excitation and detection linear polarizers (Figure ). The combination of the polarization direction of the excitation and detection polarizers in the LPRS/LPAOS measurements is expressed as “XY”. X and Y are the polarization direction of the excitation and detection polarizers, respectively, and they can both take V or H where V refers to vertical polarization; that is, the polarization direction is perpendicular to the instrument plane defined by the light source, sample holder, and the detector. H refers to horizontal polarization; that is, the direction of the linear polarizer is oriented parallel to the instrument plane (Figure ). For LPRS measurements, the excitation and emission monochromator wavelengths were kept identical during spectral acquisition. − In contrast, LPAOS analysis involved scanning the emission wavelength through three regimes: anti-Stokes-shifted (shorter than the excitation wavelength), on-resonance (identical to the excitation wavelength), and Stokes-shifted (longer than the excitation wavelength). −

1.

Schematic representation of the spectrofluorometer-based LPRS/LPAOS VV and VH spectral acquisition.

The combined UV–vis, LPRS, and LPAOS spectroscopic measurements have led to significant advancements in the fundamental understanding of photon–matter interactions including experimental identification of on-resonance fluorescence (ORF)a phenomenon in which fluorescence emission occurs at the exact excitation wavelengthand have enabled distinguishing ORF from light scattering. , Furthermore, this combined methodology has been instrumental in refining evidence-based approaches for interpreting UV–vis extinction and systematically investigating CPMI in optical spectroscopic measurements. ,,

Despite its unique ability to provide previously unattainable yet crucial insights for materials characterization, the combined UV–vis, LPRS, and LPAOS techniques suffer two critical limitations. The first challenge is low data acquisition efficiency: acquiring a complete data set requires two separate instruments and three sequential spectral acquisitions, which can take 5 min or longer when the number of wavelength data points exceeds 500. Such a poor efficiency makes real-time monitoring in dynamic systems impractical.

The second challenge is ensuring wavelength reproducibility across spectra obtained from different instruments (UV–vis spectrophotometer and fluorescence spectrophotometer) and under varied measurement conditions (LPRS/LPAOS VV and VH). − Additionally, polarization reproducibility presents a major concern in LPRS/LPAOS VV and VH spectral acquisitions. , Since the motorized monochromator and linear polarizer rely on stepper motors for precise positioning, any inaccuracies in motorized control can lead to misalignment in wavelength and polarization direction. By multitrack design, MLPS drastically improves the temporal resolution of the combined UV–vis and linearly polarized spectroscopic measurements and effectively mitigates these reproducibility issues because it enables the concurrent collection of UV–vis and linearly polarized VV/VH data under identical measurement conditions, eliminating inter-instrument and -measurement variations. A complete list of abbreviations is provided in the Supporting Information.

2. Experimental Section

2.1. Materials and Reagents

Polystyrene nanoparticles (PSNPs) with a nominal diameter of 200 nm were obtained from Polysciences. Analytical grade anthracene, ovalene, compound 610 (C610), and Rhodamine B (RhB) solid blocks used for quantification of the MLPS G-factor spectrum were purchased from Starna Scientific. Potassium permanganate (KMnO₄) was purchased from Sigma-Aldrich. meso-Tetrakis­(4-sulfonatophenyl)­porphyrin (TSPP) (Production Number: A5013) was purchased from TCI America. Nanopure water (18.2 MΩ·cm) was used in solution preparation.

2.2. Sample Preparation and Measurements

The self-assembly of TSPP was initiated according to the literature procedure. In brief, 3 mL of 6 μM TSPP in 0.01 M NaOH solution is placed in a quartz fluorescence cuvette under constant stirring (1500 rpm). TSPP assembly is initiated by rapidly adding 50 μL of 12 M HCl. Kinetic spectral acquisition begins right before HCl addition. The assembly process was monitored in situ with 3600 kinetic MLPS measurements, each with an integration time of 0.5 s and a readout time of 0.15 s. As such, the temporal resolution is 0.65 s. Notably, for faster processes, the time resolution is expected to be 0.01 s by using an electron-multiplying CCD (EMCCD). All spectra were collected at room temperature using a 1 cm × 1 cm square fused quartz fluorescence cuvette (Thorlabs) containing 3 mL of sample.

3. Results and Discussion

3.1. MLPS Instrumentation

The schematic and photograph of the MLPS instrument are shown in Figure . This instrument comprises three main modules: a free-space excitation module, a sample compartment module, and a fiber-coupled multitrack CCD spectrometer module.

2.

(A) Schematic and (B) photograph of the MLPS spectrometer. BS: beam splitter; FC: fiber collimator; PH: pinhole; CCD: charge-coupled device; LP: linear polarizer; FT: forward track; RT: reference track; XV: sideward track with vertical polarization detection; XH: sideward track with horizontal polarization detection.

The excitation module comprises a 300 W Xe lamp and a 1/8 m monochromator (Sciencetech TLS 72-X300) with adjustable slit width. The central excitation wavelength can be changed from 300 nm to 1800 nm with an accuracy of 0.3 nm. The output beam is collimated by a 50.8 mm diameter lens (Lens1) with a focal length of 100 mm (Thorlabs LA4545), descended by a periscope assembly (Thorlabs RS99), and further collimated with lens 4 to a nonpolarizing beam splitter with a transmittance/reflectance ratio of 90:10 (Thorlabs BS025). The transmitted portion (90%) passes through a linear polarizer (LP1) (Thorlabs WP25M-UB), referred to as the excitation polarizer hereafter, and then illuminates the sample.

The polarization direction of the excitation polarizer can be manually rotated between horizontal (H) and vertical (V) orientations, with the horizontal setting used to calibrate the instrument factor, i.e., the polarization-bias in the signal response or vertical (V) direction for the polarization-resolved scattering and fluorescence spectral acquisition (vide infra). The reflected portion (10%) of the light from the monochromator is focused on one fiber designed as the reference track.

The sample compartment module comprises a Peltier-based temperature-controlled cuvette holder with four optical ports, allowing the temperature to vary from −15 to 100 °C (Quantum Northwest Luma 40) with 0.1 °C precision. Both the temperature and stirring speed can be controlled and monitored with a commercial controller (Quantum Northwest TC 1).

The fiber-coupled multitrack CCD spectrometer module uses a seven-leg fiber bundle (FiberTech Optica 7 × 200 μm high hydroxyl ion fiber) for the spectral acquisition. All optical fibers are arranged in a linear configuration at the end of the fibers connected to the CCD spectrometer. Only alternative vertically arranged fibers (#1, #3, #5, and #7) are used for data acquisition to mitigate the crosstalk among the spectra collected with different tracks. Besides the reference track used for monitoring excitation intensity variations, the other three fibers used for collecting the optical signal generated from the sample solution are designed for forward track and sideward XV and XH tracks, respectively. While the forward detects light along the propagation direction of the incident photons, the sideward tracks collect photons propagating sideward from the direction of the incident light. Sideward XV and XH tracks differ in the polarization direction (V or H) of the linear polarizers placed in front of their respective fiber coupler. X represents the direction of the excitation linear polarizer that can be H or V depending on the measurement needs.

The photons collected with the four fiber tracks are dispersed with a Czerny-Turner spectrograph with a 328 mm focal length (Andor KYMERA-328I) and equipped with a 150 grooves/mm grating and a blaze wavelength of 500 nm. The dispersed light is detected by a deep-cooled CCD camera with an image area of 26.6 × 6.6 mm or 1024 × 256 pixels (Andor iDus 420A-BEX2-DD) and an operation temperature of −80 °C. When operating with a central wavelength at 650 nm, this CCD spectrometer has a wavelength coverage of 540 nm (from 380 to 920 nm) or an average of 0.52 nm/pixel.

The forward and sideward XV and sideward XH tracks are arranged in a T-format configuration. A small pinhole (∼1.5 mm) and a set of absorptive neutral density filters with an optical density (OD) of 1.0, 3.0, and 5.0 (Thorlabs NE10B-A, NE20B-A, NE50B-A) are mounted in a cage filter wheel (Thorlabs CFW6/M) in front of the fiber port for the forward track. The pinhole is critical for (a) mitigating the interference of forward scattering light on UV–vis extinction measurements and (b) reducing the signal saturation of forward spectra at the excitation wavelength.

The two sideward signals are collected using 50.8 mm diameter lenses (Lens2 and Lens3) with a focal length of 60 mm (Thorlabs LA4464) and then pass through a linear polarizer (Thorlabs WP25M-UB) polarized horizontally or vertically to the instrument plane defined by the incident light, the sample holder, and the CCD spectrometer. Four high numerical aperture (NA = 0.54) multimode fiber collimators (Thorlabs F950SMA-A) are used for coupling the light into the fibers.

3.2. MLPS Characterization

CCD Binning

Experimental procedures and data for determination of the CCD pixel binning for different spectral tracks are detailed in the Supporting Information. A binning size of nine CCD-pixel rows per spectral track was employed in the MLPS measurements (Figure ), which is determined by maximizing the spectral intensity of each intended track while minimizing the spectral crosstalk between signals collected from different optical fibers. The maximum crosstalk under this configuration is less than 0.2% (Figure ), as evaluated with the excitation wavelength centered at 550 nm. This level of crosstalk is negligible for the applications explored in this work. Further mitigation of crosstalk can be achieved by reducing the spectral binning sizes, increasing the distance of fibers used for different spectral tracks, and utilizing optical fibers with smaller core diameters.

3.

(A) Combined normalized vertical pixel intensity when individual tracks are used for light collection. (B) Degree of crosstalk. Only signals in the shadowed pixel rows are binned for the data acquisition

CCD Dark Signal

The dark signal of the CCD detector was characterized by measuring the output signal from each pixel with no incident light on the detector. At an operational temperature of −80 °C, the measured dark current remained within 492 to 496 counts for each spectral wavelength. This dark current is independent of the integration times from 0.05 to 2 s and the wavelength (Figure S1), indicating it is governed primarily by CCD reset bias and readout noise. To mitigate the interference of dark current in spectral analysis, a constant dark signal of 492 counts was subtracted from the raw experimental data prior to any data analysis.

Wavelength Calibration and Reproducibility

Ensuring high wavelength reproducibility across spectral tracks is crucial for achieving accurate concurrent multispectral analysis in MLPS. A mercury–argon light source (HG-2, Ocean Optics) is used to calibrate the spectrometer wavelength. Optimizing the positioning of the entrance end of the fiber bundle through a trial-and-error approach resulted in a wavelength reproducibility of ±0.5 nm among the MLPS spectral tracks (Figure A and B), as evaluated using a set of single-band wavelengths ranging from 400 to 700 nm. Such wavelength mismatch is smaller than the pixel wavelength resolution (0.52 nm/pixel), highlighting the high precision of the fiber bundle alignments. This level of wavelength reproducibility is sufficient for most intended MLPS applications, as the spectral bandwidth of UV–vis, scattering, and emission spectral features in photoactive materials are typically much broader than, for example, 10 nm. The high wavelength reproducibility allows the same wavelength calibration curve (Figure C) to be applied for four-track spectra, streamlining the data analysis.

$$P(\lambda )=G(\lambda )\frac{I_{VH}(\lambda )}{I_{VV}(\lambda )}$$ 1

$$G(\lambda )=\frac{I_{HV}(\lambda )}{I_{HH}(\lambda )}$$ 2

4.

(A and B) Wavelength reproducibility among the four MLPS tracks evaluated with monochromatic light with central wavelengths from 400 to 700 nm. The curves in red, blue, green, and black correspond to the reference, forward, sideward XV, and sideward XH tracks, respectively. The spectral intensity is arbitrarily scaled to facilitate the comparison. (C) Detected wavelength calibrations as a function of excitation wavelength.

G-Factor Spectrum

The sideward XH and XV tracks in MLPS are designed to detect polarization-resolved scattering and fluorescence intensities. The data acquired from individual tracks are valuable for probing the sample’s scattering and fluorescence intensity, and the ratio between VH vs VV enables quantification of the scattering and fluorescence depolarizations P(λ) (eq ). These two parameters are widely used in materials’ characterization, as they provide crucial information on molecular orientation, structural anisotropy, and dynamic interactions.

Normalizing the signal responsivity of the two sideward tracks, which differ in detection polarizations (H or V), is needed to ensure the accuracy in calculating scattering and fluorescence depolarization. The signal normalization curve, MLPS G-factor spectrum, is defined with eq and experimentally quantified using the same method developed for determining the G-factor spectrum (eq ) for spectrofluorometer-based linearly polarized spectroscopic measurements. Briefly, a set of linearly polarized fluorescence (HV and HH) spectra is acquired for multiple fluorescence solutions with overlapping emission wavelengths, which collectively cover the wavelength range for the intended application. Since excitation light is H-polarized, the intensities of H- and V-polarized fluorescence light detected 90 degrees relative to the incident beam should be theoretically the same. ,, As such, any observed signal discrepancy between the sideward HH and HV spectra must be attributed to the difference in signal responsivity between the two MLPS sidetracks, corresponding to V and H detection polarizations, respectively.

The signal responses between the MLPS HV and HH signals are very similar (Figure A–E), leading to a near-unity MLPS G-factor across the entire wavelength region (Figure F). In contrast, the G-factor value of the conventional spectrofluorometer, which utilizes free-space excitation and detection and has been extensively used in our previous linearly polarized spectroscopic measurements, varies significantly, ranging from 1.5 to 0.5 depending on the excitation wavelength. ,, The stark difference in the G-factor spectra between the fiber-coupled MLPS instrument and free-space commercial spectrofluorometer is likely due to the depolarization effects of the multimode optical fibers used in the MLPS setup. , While the gratings in the spectrofluorometer’s detection monochromator and the MLPS CCD spectrometer likely have a similar polarization bias, the 2 m multimode optical fibers in the MLPS system effectively depolarize the light before it reaches the CCD spectrometer. Regardless of its underlying causes, the near-unity MLPS G-factor spectrum enables direct examination of the degree of the scattering and fluorescence depolarization or anisotropy.

$$E(\lambda ,t)=-\log \left(\frac{I_{ft,Sample}(\lambda )/I_{ref,Sample}(\lambda )}{I_{ft,Solvent}(\lambda )/I_{ref,solvent}(\lambda )}\right)$$ 3

5.

(A–E) Linearly polarized fluorescence (black) HV and (red) HH emission spectra collected using the two MLPS sidetracks. (F) MLPS G-factor spectrum.

Linear Dynamic Range (LDR) for UV–Vis Extinction

The forward track is designed for quantification of solution UV–vis extinction (eq ). LDR of MLPS-based UV–vis extinction quantification was evaluated with two types of solutions, KMnO₄ as approximately pure absorbers (Figure A and C) and PSNPs as approximately pure light scatterers (Figure B and D). LDR of commercial UV–vis spectrophotometers is typically evaluated using only solutions that are pure absorbers. However, due to the inevitable interference of forward light scattering, the upper limit of instrument LDR in terms of scattering extinction is invariably lower than that in terms of absorption extinction, or absorbance. As an example, for the commercial UV–vis spectrophotometer used in this work, its upper LDR limit of the KMnO₄ absorbance is about 5, while that for PSNPs depends strongly on the particle sizes and becomes less than 3 for the same PSNPs (diameter 200 nm) used in this work. Such a drastic difference indicates the significance of the interference of forward scattering light.

6.

Forward intensity spectra of (A) KMnO₄ and (B) PSNP solution. Spectra in black are the solvent control. Correlation between (C) MLPS UV–vis absorbance vs KMnO₄ absorbance and (D) MLPS UV–vis scattering extinction and the theoretical scattering extinction. The MLPS integration time is 0.5 s, and the central excitation wavelength is 490 nm. The theoretical absorbance and scattering are deduced from measurements with a commercial UV–vis spectrophotometer.

The upper LDR limit of MLPS-based UV–vis measurement is between 4 and 5 in terms of KMnO₄ absorbance, while that for PNSP scattering extinction is below 4. The fact that MLPS has a more-or-less similar LDR for absorption and scattering extinction is due to the small, ∼1.5 mm, pinhole used in the design. Without this pinhole, the MLPS LDR for KMnO₄ remains approximately the same, while that for the PSNP scattering extinction is reduced drastically to less than 0.5 (Figure S2). This observation is consistent with theoretical modeling that predicts that the larger the detector collection angle, the lower the upper limit of the UV–vis LDR for scattering solutions.

The comparable LDR of pinhole-equipped MLPS for UV–vis absorption and scattering extinction is particularly critical for investigating the assembly and disassembly of photoactive materials. For instance, dissolved molecular and dispersed ultrasmall nanoparticle chromophores function primarily as pure absorbers. In contrast, their assembled states serve as both absorbers and scatterers, with the ratio of scattering to absorption intensity varying depending on the dynamics of assembly and disassembly processes. To ensure the reliability of UV–vis extinction measured with the current MLPS instrument, the sample design should be such that its kinetic UV–vis extinction remains below 3 during the dynamic physicochemical processes where temporal variation of solution scattering and absorption extinction ratio is unknown.

The MLPS UV–vis LDR discussed above was determined for kinetic studies that require a fixed short integration time (0.5 s, in this work). For steady-state solutions, MLPS UV–vis LDR can be extended by establishing the excellent linear relationship between forward spectral intensity and CCD integration time (Figure S3). To demonstrate this capability, KMnO₄ was used as a model absorber; the upper LDR limit for MLPS UV–vis absorbance was successfully extended to 5.6 by increasing the integration time from 0.5 s for diluted solutions to 100 s for the most concentrated KMnO₄ (Figure S4). By utilizing longer integration times, MLPS can significantly extend the LDR limit for UV–vis absorbance, making it advantageous for analyzing highly concentrated samples and other materials that quickly exceed the LDR of conventional spectroscopic instruments.

3.3. Processing and Analysis of Kinetic MLPS Data

Time-Domain Cosmic Spike Removal

A universal complication with CCD-based spectral acquisition is cosmic spike interference, which appears as random spectral peaks caused by electromagnetic radiation, such as γ rays originating from space. Figure A shows the as-acquired kinetic sideward VH spectra obtained from TSPP assembly solutions triggered by addition of HCl. Multiple cosmic spikes are present in the kinetic data, including two particularly intense ones that dominate the spectra.

7.

(A) As-acquired kinetic MLPS VH spectra, (B) an example of the time-domain cosmic spike removal, and (C) cosmic-spike-removed MLPS VH spectra. Insets show the spectral dips. (D) The data after cosmic spike removal followed by spectral dip correction.

While manual identification and removal of cosmic spikes may be feasible for small data sets, it becomes challenging for large-scale spectral acquisitions and unsuitable for dynamic systems where spectral properties evolve over time. For steady-state applications, such as Raman imaging, where it involves multiple CCD spectra for the same sample, an upper-bound spectrum method can be applied for automatic cosmic spike identification and removal. Since the cosmic events are random in time and space (across CCD pixels), the cosmic spikes can be statistically identified as outliers in the spectral series. Once detected, the identified cosmic spike intensity is replaced with an upper-bound intensity estimated based on the statistics of the spectra taken with the same sample. However, this approach is unsuitable for dynamic systems, where the sample optical property changes over time.

In this work, a MATLAB program (Figure S5) is developed for cosmic spike removal in the TSPP assembly kinetic spectra. The method is termed as time-domain cosmic spike removal, as it involves two assumptions for the time-course of signal intensity at specific wavelength pixels. The first is that the probability of a cosmic spike occurring at a specific pixel across three consecutive temporal points is negligibly small. The second assumption is that the spectral intensity changes in three consecutive temporal points can be approximated as linear. Based on these assumptions, a temporal signal is deemed to be a cosmic spike if its dark-current-corrected intensity is 25% higher than the average signal of its two adjacent temporal points. In this case, the spike intensity is replaced with the average of the neighboring temporal points. As shown in Figure B and C, this cosmic spike removal method effectively eliminates all detected cosmic spikes while preserving the integrity of the kinetic spectral data. Moreover, it takes less than 30 s to process and remove the cosmic spikes from all 3600 spectra.

The first assumption is valid for most kinetic MLPS measurements, provided the spectral integration time is short (e.g., <10 s), as cosmic ray radiation is a relatively rare event. However, the validity of the second assumption is sample-dependent and more critically influenced by the temporal resolutions of the kinetic MLPS measurements. High temporal resolution is required for dynamic systems, where rapid changes in photophysical properties occur.

Wavelength-Domain Correction of Spectral Dips

Besides cosmic spikes, another artifact in as-acquired MLPS data is spectral dips, which refer to the lower-than-expected spectral intensity (Figure C). Unlike cosmic spikes, which appear randomly in time, spectral wavelength, or CCD pixels, spectra dips are wavelength- or pixel-specific and consistently occur at the same wavelength in every kinetic spectrum.

Empirical observations indicate that the maximum dip intensity is approximately 30 counts, and the dip pattern and intensity are independent of spectral integration time and the sample solution (Figure S6). These observations exclude the possibility of quantum efficiency between neighboring pixels. Otherwise, the dip magnitude would be proportional to the spectral integration time. Despite efforts to understand mechanisms including communication with the CCD vendor, the origin of these fixed-pattern and -magnitude artifacts remains unclear. Since the spectral dips are within the shot noise when the CCD signal is high (e.g., ≥10,000), they are essentially invisible in spectra with high intensities. For low-intensity measurements, such as the sidetrack MLPS spectra obtained with high temporal resolution, it is essential to remove the spectral dips before performing data analysis.

A MATLAB program for wavelength-domain correction of the spectral dips is developed by the empirical observation that most spectral dips are separated by one or more CCD wavelength pixels (Figure S7). This software identifies the spectral feature as a dip if its intensity is more than 25% lower than the average intensity of its neighboring wavelength pixels (one on each side). The effectiveness of the spectral dip removal is demonstrated in the zoomed-in kinetic spectra shown in Figure D. Note that as the spectral dips are fixed-wavelength artifacts with a relatively constant temporal magnitude, the presence of the dips has no effect on the temporal-domain cosmic spike identification and removal. However, cosmic spikes can complicate the wavelength-domain dip correction. Therefore, cosmic spike removal should be performed prior to spectral dip correction to ensure reliable spectral preprocessing.

Kinetic MLPS of TSSP Assembly

The versatility and multispectral capabilities of the MLPS instrument are demonstrated with in situ monitoring of TSPP self-assembly using two excitation wavelengths: 490 and 665 nm (Figure ). The reference spectra for these kinetic MLPS measurements demonstrate excellent excitation power stability (Figure S8). These two excitation wavelengths are chosen based on TSPP optical properties under these excitation conditions. The 490 nm peak in UV–vis and resonance synchronous spectra is a characteristic spectral feature of TSPP assembly formation, as TSPP assemblies exhibit strong light scattering at this wavelength. − However, the absorption and fluorescence activities of TSPP assemblies at this excitation wavelength have not been systematically characterized, especially in TSPP assembly processes. Moreover, the TSPP assembly exhibits UV–vis activities at 665 nm. However, the mechanistic origin of UV–vis extinction at this wavelengthabsorption, scattering, or bothremains unknown.

8.

In situ kinetic MLPS characterization of the TSPP assembly with the excitation wavelength of (left) 490 and (right) 665 nm. The data obtained in the reference track are shown in the Supporting Information. (A and B) Forward spectra, (C and D) sideward VV spectra, and (E and F) sideward VH spectra. The insets are the zoomed-in features in the plotted wavelength regions. Red, solvent; dark blue, before HCl addition; light blue and pink, the first and last kinetic spectra; gray, the ones taken in between.

The MLPS data provide previously unattainable insights that are essential for evidence-based spectral data analysis. It unequivocally demonstrates that TSPP assemblies are simultaneous light absorbers and scatterers but not an ORF emitter under 490 nm excitation, while they concurrently absorb, scatter, and emit, all at 650 nm excitation. Therefore, their UV–vis extinction should be interpreted as the sum of absorption and scattering extinction. The absorption activities at 490 nm wavelength are evident by the Stokes-shifted fluorescence (SSF) observed in the 600 to 750 nm spectral region, while its scattering activity is confirmed by the strong sideward VV and VH intensities at 490 nm (Figure C, E). ORF activity at 490 nm is excluded by the absence of anti-Stokes-shifted fluorescence (ASSF) in the VV and VH spectra under 490 nm excitation, as ORF invariably appears together with ASSF and SSF.

In contrast, at 665 nm excitation, MLPS VV and VH spectra exhibit ASSF, ORF, and SSF emission, confirming that the TSPP assembly is both a light absorber and an ORF emitter at this wavelength. The 650 nm peak in the TSPP VV spectra indicates that TSPP is also a light scatterer at this wavelength (Figure D). Traditionally, ORF is often misinterpreted as a sample’s light scattering because both occur at the excitation wavelength. The data obtained with 665 nm excitation reveal that the ORF and scattering have comparable intensities in the VV spectra (Figure D), but the VH signal at the excitation wavelength is dominated by the ORF with no significant scattering contribution (Figure F).

The kinetic MLPS data enable simultaneous quantification of TSPP UV–vis extinction at the excitation wavelengths (Figure A and B), polarization-resolved scattering (Figure C and D), and fluorescence emission intensities (Figure E and F). Additionally, it allows for the calculation of scattering and fluorescence depolarizations (Figure G and H) by using eq , all with a temporal resolution of 0.65 s. Obtaining similar information using combined UV–vis spectrophotometric and spectrofluorometric measurements would require at least 5 min at each temporal point, showing that the MLPS spectrometer offers a more than 400-fold improvement in temporal resolution. While the correlation between the structural evolution of TSPP assemblies and the kinetics of MLPS spectral features requires further investigation, the proof-of-concept application presented here demonstrates MLPS’s potential to revolutionize optical spectroscopic studies of dynamic systems, where existing optical tools remain inadequate.

9.

Time-courses of (A and B) UV–vis extinction, (C and D) scattering VV and VH intensity, (E and F) VV and VH fluorescence emission at 718 nm, and (G, H) the scattering and fluorescence depolarization deduced from the data presented in Figure .

4. Conclusion

This work introduces and characterizes a multitrack linearly polarized spectrometer (MLPS), a CCD-based optical system designed for simultaneous kinetic UV–vis, scattering, and photoluminescence measurements with high temporal resolution. Unlike conventional optical spectroscopic tools that often deliver fragmented data, MLPS enables the concurrent quantification of UV–vis extinction, polarization-resolved scattering, and fluorescence intensities, as well as scattering and fluorescence depolarizations, thereby overcoming limitations in studying dynamic systems. By integrating fiber-coupled spectral acquisition, polarization-resolved detection, and a multitrack CCD sensor, MLPS allows for a more comprehensive and accurate spectral analysis than was previously possible. The MLPS system’s performance was evaluated through a series of benchmarking experiments, comparing it to commercial UV–vis spectrophotometers and spectrofluorometers. The results demonstrate that MLPS achieves wavelength reproducibility better than 0.5 nm across the 400–700 nm range, which is sufficient precision for most photoactive material studies. Additionally, the system shows near-unity responsivity between its two sideward polarization tracks, significantly minimizing polarization bias and enabling accurate scattering and fluorescence depolarization measurements. The application of a newly developed cosmic spike removal algorithm further enhances the reliability of kinetic spectral acquisitions, making the MLPS particularly suited for studying rapid photophysical processes. A key demonstration of MLPS’s capabilities involved the real-time monitoring of porphyrin self-assembly, where its high temporal resolution (subsecond) provided unprecedented insights into the structural and optical evolution of dynamic molecular systems. Beyond its demonstrated applications, the MLPS framework has the potential to significantly advance research in nanomaterials, biointerfaces, and photophysics, where precise, time-resolved optical characterization is critical. Future optimizations may include extending the spectral range, improving the temporal resolution, and integrating real-time corrections for spectral distortions. Overall, this study establishes MLPS as a powerful and versatile tool for next-generation optical spectroscopy, paving the way for more accurate and holistic investigations of dynamic material systems. Additionally, the instrument characterization and data preprocessing techniques presented here provide valuable insights into the development of future multitrack CCD-based spectrometers.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmeasuresciau.5c00022.

• Additional details on CCD characterization, MLPS measurement performance, and data processing methodologies; a description of CCD binning strategies used to optimize spectral resolution and signal-to-noise ratio, along with an evaluation of CCD dark signal characteristics, highlighting baseline intensity levels and integration time independence; MLPS LDR comparison measurements examined without using a pinhole, demonstrating the impact on collection-detected signal; the linear relationship between signal intensity and CCD integration time; a MATLAB-based algorithm for the automated removal of cosmic spikes and spectral dips, along with an investigation into the intensity variations and pixel-level separation of spectral dips; reference spectra for kinetic MLPS measurements to ensure reproducibility and consistency in spectral analysis (PDF)

Rongjing Yan: Data acquisition, instrument performance experiments, data analysis, writing – review – editing. Pathum Wathudura: TSPP kinetic experiments, instrument troubleshooting, writing – review. Qiang Hao: Instrument construction, writing – review. Max Wamsley: Initial instrument setup. Willard E. Collier: Writing – review. Dongmao Zhang: Research project supervision, conceptualization, funding acquisition, instrument design, supervision, writing – original draft, review, and editing.

Δ.

R.Y. and Q.H.: Equal contribution. CRediT: Rongjing Yan investigation, software, validation, visualization, writing - original draft, writing - review & editing; Qiang Hao investigation, methodology, writing - original draft, writing - review & editing; Pathum Wathudura investigation, software; Max Wamsley investigation, methodology, software; Willard E. Collier writing - review & editing; Dongmao Zhang conceptualization, formal analysis, funding acquisition, investigation, methodology, project administration, resources, software, supervision, validation, visualization, writing - original draft, writing - review & editing.

The authors declare no competing financial interest.