Using a 3D Silicon Micro-Channel Device and Raman Spectroscopy for the Analysis of Whole Blood and Abnormal Blood

1. Introduction

Conventional blood tests are performed with optical microscopes, which are time-consuming and may lesser the accuracy of blood sample analysis. Among the emerging laser-based blood testing, Raman spectroscopy is one of the methods that involves nonlinear scattering. This method is adopted in a wide range of blood tests because of its various advantages, such as a small volume of blood for testing, rapid analysis, and high-precision result interpretation. These factors simplify the process of collecting and evaluating blood-related data. The effectiveness of Raman spectroscopy in clinical applications has been verified [1,2].

In Raman spectroscopy, the wavenumber difference between incident and scattered light is compared to determine the changes in their energy levels and the sample’s lattice vibration characteristics. These observations can be used to analyze the composition and structure of the sample. Therefore, the Raman peak values can indicate a sample’s molecular structure [3,4].

Raman spectroscopy can be applied in blood testing to determine the presence of certain disease in individuals. The Raman characteristic peak positions and amplitudes exhibit distinct differences between infected and uninfected blood samples. Therefore, Raman spectroscopy serves as a valuable tool for rapid pathogens screening and identification. The prevalence of emerging viruses, such as COVID-19, will eventually overlap with that of influenza virus infection, and this trend has necessitated reliable, rapid and disease screening [5]. Moreover, the acquisition of blood information through Raman spectroscopy enables the rapid and precise diagnosis of clinical diseases, such as liver disease, diabetes, cancer, allergies, genetic disorders, anemia, and leukemia [6,7,8].

In the present study, we implemented 3D Silicon Micro-Channel Device (3D SMCD) to introduced blood and calcium coagulants for mixing. The primary Raman peaks in whole blood was attributed to hemoglobin, while other Raman peaks corresponded to amino acids and other components [9]. In contrast, the primary Raman peaks of plasma and serum corresponded to β-carotene [10]. Moreover, we conducted a prothrombin time (PT) experiment to compare the Raman spectroscopy results of blood samples with and without calcium-containing coagulant. Clinically, coagulation or clotting, in which blood turns from liquid to non-flowing gel, is a critical process in hemostasis [11,12]. This phenomenon can be confirmed using a transmission Raman spectrometer in the mixing zone of the 3D SMCD [13]. The primary objective is to determine whether patients’ intrinsic and common coagulation pathways were functioning normally.

2. Sample Preparation for 3D SMCD

2.1 Fabrication of 3D SMCD

The 3D SMCD (sandwich structure-Poly(methyl methacrylate), PMMA/Silicon/PMMA) heterogeneous structure [13] was fabricated using a low-temperature bonding technique. The blood mixing area was bonded to the lower layer of the 3D SMCD (Figure 2(A)). This helps prevent product breakage or blood leakage. This 3D SMCD design of the blood-specific automated mixing system pro-vides external systems for blood testing. As evidenced by the Raman spectroscopy signals obtained in this study’s experimental results, successful acquisition of biological component information was collected.

The results demonstrate the successful development of a novel 3D SMCD, integrated with a Raman spectroscopy system for blood clotting detection. We fabricated the 3D SMCD, featuring a sandwich structure composed of PMMA (Polymethyl Methacrylate)/silicon/PMMA, utilizing a low-temperature bonding process to achieve the bonding of heterogeneous materials while eliminating internal stresses. This helps prevent product breakage or blood leakage. We successfully achieved a maximum bonding surface energy value of 841 mJ/m², ensuring no leakage during blood measurements. In the lower mixing zone of the 3D silicon-based microfluidic device, we integrated the Raman spectroscopy system to obtain characteristic spectra from Raman spectroscopy signals for blood clotting measurements, as shown in Figure 1.

2.2. Blood Sample Preparation

All blood samples in this study were obtained with the approval of the institutional review board from blood donors at the National Cheng Kung University Hospital in Taiwan. Three types of samples were extracted for each blood specimen. None of the normal samples used in this study were subjected to any chemical or biological treatment. Blood samples were drawn using a syringe, dropped on an upper hole of the 3D SMCD (100 μL). All the experiments were conducted on the blood samples at room temperature (RT). Whole blood, which was obtained from healthy individuals, typically had a PT of 10.0–12.5 s. By comparison, the coagulation factor required to bind to calcium (Ca⁺⁺) ions in the blood, called factor IV, is known as abnormal blood, which was obtained from patients receiving anticoagulant drugs, had a clotting time of longer than 12.5 s. For experimental protocol, a calcium-containing coagulant had to be mixed with blood in a ratio of 2.8:1. To meet the requirements of clinical practice, the prepared whole blood and plasma samples were also examined as shown in Figure 2.

Figure 1. Trend curve of plasma power and bonding energy of 3D SMCD.

Figure 1. Trend curve of plasma power and bonding energy of 3D SMCD.

Figure 2. Preparation of blood samples using 3D SMCD, (A) 3D SMCD, (B) FE-SEM image of the mixing zone of the microchannel, (C) Blood sample in test tube.

Figure 2. Preparation of blood samples using 3D SMCD, (A) 3D SMCD, (B) FE-SEM image of the mixing zone of the microchannel, (C) Blood sample in test tube.

3. Raman Spectroscopy

Raman spectroscopy was performed at RT using a custom-made micro-Raman spectroscopy system consisting of a spectrometer (Princeton Instruments Acton Spectra Pro 2500i), a liquid nitrogen cooled CCD detector (Princeton Instruments Acton 7508-0002), four objective lenses from 20x to 100x, and a solid-state green laser (wavelength of 532 nm) as its excitation light source with the laser output power of 2 mW. To avoid laser heating damage on blood samples, each Raman spectrum was recorded with the integration of 3 s, 5 times. The spectral range was set between 400 and 2000 cm⁻¹ as shown in Figure 3. The Raman characteristic peak position was calibrated with measurements of a Si(100) substrate under the same conditions, and the background intensity was subtracted. In addition, the Raman spectra are normalized to the characteristic peak of whole blood at 1562 cm^-1 [14], as shown in Figure 3.

4. Results and Discussion

4.1. Raman Spectrum for Whole Blood

On the basis of literature review and matching inquiries in the Spectral Database for Organic Compounds [15], we found that the most abundant component in the whole blood is hemoglobin [16], followed by amino acid, lactic acid, lactate, and acetate. Hemoglobin exhibits characteristic peaks at 690, 989, 1015, 1182, 1233, 1315, and 1562–1649 cm⁻¹. Hemoglobin constitutes 95% of the weight of a red blood cell and affects the oxygen-carrying capacity [17]. In clinical practice, hemoglobin level can serve as a standard for anemia detection [18], and the severity of COVID-19 disease can be predicted using blood oxygen level [5]. Amino acid is a small molecule [19] which exhibited characteristic peaks at 770 and 1348 cm⁻¹. Moreover, lactic acid and lactate exhibited characteristic peaks at 1092 and 1144 cm⁻¹, respectively. Lactic acid is a metabolic product of carbohydrates, and blood lactic acid levels can serve as a standard for the diagnosis of lactic acidosis in clinical practice [20]. Lactate is a salt produced when lactic acid releases a hydron and reacts with a positively charged ion like sodium or potassium. An excessive lactate level is recognized as a potential cause of acidemia (acidosis) [21]. The main characteristic peak of acetate was located at 1441 cm⁻¹ as shown in Figure 4. Acetate exists in animal tissues, excreta, and blood in the form of free acid. Table 1 displays the characteristic Raman peaks and vibrational modes of the components found in whole blood. [22]. These components may have lower concentrations in blood or undergo changes post-coagulation, resulting in weaker and less distinguishable peaks. Furthermore, the spectral range between 600 cm^-1 and 800 cm^-1 may encompass characteristic peaks of other biological information such as hormones, antibodies, and enzymes. The Raman detection results revealed that due to structural changes in platelets during low-temperature storage and potential lipid loss, some peaks were weaker and less distinguishable.

4.2. Comparison of the Raman Spectra of Whole Blood and Abnormal Blood Function of Factor IV

By comparing the Raman spectra between normal whole blood and blood with abnormal Factor IV function, the critical difference becomes apparent within the spectral range of 1562 to 1651 cm^-1, as shown in Figure 5. The whole blood sample exhibited characteristic peaks at 1562 and 1613 cm⁻¹, whereas the abnormal blood sample exhibited a characteristic peak at 1574 cm⁻¹. This result suggests that the coagulation of hemoglobin was easier in whole blood than in abnormal blood because of the shorter time period for clotting in normal blood. Accordingly, the characteristic peaks of aromatic ring breathing, C==N antisymmetric stretching, and pyrrole ring stretching were either absent or shifted in peak position in the Raman spectrum of abnormal blood function of Factor IV [23]. Table 2 presents the characteristic Raman shifts and corresponding components for whole blood and abnormal blood function of Factor IV.

4.3. Comparison of the Raman Spectra of Abnormal Blood (Factor IV) with and without Coagulant

A comparative study of abnormal blood (Factor IV) with and without a coagulant was conducted, as shown in Figure 6. The results indicate the intrinsic pathway can be initiated by events that take place within the lumen of blood vessels. The intrinsic pathway requires only elements (coagulation factors, Ca⁺⁺, platelet surface, etc.) found within or intrinsic to the vascular system. In the comparison between abnormal blood with and without coagulant, the abnormal blood (Factor IV) with coagulant had lower intensities at the characteristic peaks of 770, 1011, 1185, 1318, 1356, and 1384 cm⁻¹ and higher intensities at 988 and 1577 cm⁻¹. Such a change in intensity, without a shift in peak position, is attributed to a (significant) change in the clotting time of hemoglobin. The addition of coagulant shortens the clotting time, and therefore, the abnormal blood (Factor IV), originally having a long time period of clotting, became fully coagulated after the addition of coagulant. Table 3 presents the major Raman shifts and corresponding components for normal blood without coagulant and abnormal blood (Factor IV) with coagulant.

On the other hand, comparison of the Raman spectra of whole blood with coagulant and abnormal blood (Factor IV) with coagulant was performed and presented in Figure 7. The characteristic peaks of abnormal blood (Factor IV) with coagulant were similar to those of whole blood with coagulant. Thus, the addition of coagulant to abnormal blood (Factor IV) resulted in the wavenumber and intensities of its characteristic peaks gradually approaching those of the characteristic peaks of whole blood with coagulant. This indicates that the coagulation cascades were commonly shared among individuals with either whole blood or abnormal blood (Factor IV) samples, but the time required for complete coagulation varied. In addition, the presence of coagulant effectively shortened the time period of clotting for abnormal blood (Factor IV) and caused full coagulation.

Therefore, in the experiment, we placed blood samples in the 3D SMCD and treated them for approximately 15 seconds. Additionally, we measured PT, a method for assessing blood clotting ability, by adding Ca⁺⁺ to the microfluidic device. To achieve rapid clotting, we automatically circulated blood within the mixing area, completing the process in about 3 seconds. These components may have lower concentrations in blood or undergo changes post-coagulation, resulting in weaker and less distinguishable peaks. Furthermore, the spectral range between 800cm^-1 and 900 cm^-1 may encompass characteristic peaks of other biological information such as hormones, antibodies, and enzymes. The Raman detection results revealed that, due to structural changes in platelets during low-temperature storage and potential lipid loss, some peaks were weaker and less distinguishable.

4.4. Comparison of the Raman Spectra of Whole Blood, Plasma, and Serum

The whole blood is the normal blood that contains anticoagulants to inhibit coagulation. The characteristic peaks of whole blood had lower intensities than those of whole blood because of the lack of coagulation in whole blood, which resulted in its molecules being unable to achieve resonance. However, the overall spectrum obtained for normal blood and whole blood were similar (Figure 8). By contrast, plasma and serum exhibited characteristic peaks only at 1016, 1164, and 1526 cm⁻¹ because only some small molecules remained after the removal of blood cells. A study suggested that the characteristic peaks of plasma and serum correspond to β-carotene [24], which only exhibits molecular resonance at low frequencies. In clinical application, changes in the characteristic β-carotene peaks in the Raman spectra of serum and plasma can be used to diagnose certain diseases, such as thyroid disease and chronic renal failure. Researchers have recommended using the Raman shift of β-carotene for diagnosing chronic renal failure and thyroid disease [25]. Table 4, Table 5 and Table 6 present the major Raman shifts and corresponding components for whole blood [21], plasma [25], and serum [26].

Raman shift (cm-1)	Vibrational blood mode	Component
1016	C-H bending	β-Carotene
1164	C-C stretching	β-Carotene
1526	C-C stretching	β-Carotene

5. Conclusions

We conducted a study utilizing a Raman spectrometer (excitation wavelength- 532 nm) to analyze blood components using a 3D silicon-based microfluidic device with both upper and lower mixing zones [27]. The combination of Raman spectroscopy with the microfluidic device allowed for more precise analysis and information retrieval of blood components. The method proved to be effective in measuring Raman spectral signals of whole human blood, offering cost-effectiveness and simple operation compared to high-precision confocal micro-Raman systems. Additionally, short-wavelength Raman scattering spectroscopy was employed to detect samples of whole blood, abnormal blood, plasma, and serum. Analysis of the biological characteristics detected in each sample type yielded several key conclusions and new findings. Firstly, the addition of a coagulant gradually made changes in the intensities and peak position of the characteristic peaks of abnormal blood resembling those of normal blood with coagulant. This behavior suggests that blood clotting affects the locations and intensities of characteristic peaks caused by hemoglobin. Secondly, the results of abnormal blood without coagulant differed from those of whole blood without coagulant, with strong differences at low-frequency vibration modes serving as potential diagnostic fingerprints. Lastly, in serum and plasma, hemoglobin is not the main molecule that resonates in Raman spectroscopy, but β-carotene exhibits molecular resonance at low frequencies, indicating its potential importance in medical testing.