1. Introduction

Raman spectroscopy is a fundamental spectroscopic technique, which is used to observe the molecular characteristic vibrational and rotational modes. Since those modes could be treated as the unique fingerprint of the analyte, it is widely used to identify chemical components and molecules [1,2]. However, this technique has perspicuous limitations in real applications on account of weak Raman scattering efficiency [3]. Many efforts have been made to enhance Raman signal, such as Surface Enhanced Raman Spectroscopy (SERS), Tip Enhanced Raman Spectroscopy (TERS), etc. On such grounds, Raman signal enhancement factor has been reported to be several orders of magnitude compared to normal Raman scattering [4,5]. However, for most of case, the volume of light-sample interaction is very small in most of the cases, which is proportional to Raman signal intensity. Besides, it is very difficult to clean up the molecules firmly adhered on the metal surface, and tiny impurities or contaminated metal surface can completely affect Raman signal. As a result, these devices cannot be reusable and are not suitable for online detection.

For online detection with unconstrained quantity of analyte, one effective way to enhance Raman signal is to increase the volume of the sample interacting with pump laser [6]. In this manner, the total Raman signal could be nearly proportional to the volume of light-sample interaction if the transmission loss is very low. As such Raman signal enhancement mechanism is based on the increasing of sample volume, it can be named as volumetric enhancement of Raman scattering (VERS). This method gives rise to its application for fast flow online detection, which has recently drawn much attention on optical sensors by confining both the sample and pump laser in the hollow-core fibers [7,8]. Low-loss hollow-core waveguide could not only be the flow channel of sample, but also extend the hot spot at the focus to a long distance and thus supports a very large volume of light-sample interaction. Furthermore, Raman signal contribution of different components is proportional to the corresponding volume ratio in the mixed sample and small amount of impurities or contaminations that may attached on the surface will have little influence on the signal.

For these reasons, hollow-core photonic crystal fibers (HCPCF) [9–13] and the metal-lined hollow-core fibers (MLHCF) [14–18] are widely used. HCPCF have been famed as an excellent “Raman signal enhancer” (up to 3 orders of magnitude) and achieved high detection sensitivity [9,10], which can guide light of certain wavelengths with low loss in a long distance. Several meters HCPCF have been used to increase the sample volume and enhance laser-analyte interaction [11,12]. However, HCPCF also has some disadvantages, albeit with remarkable sensitivity, especially not suitable for fast online detection. First of all, it takes a longer time to inject the sample into HCPCF and high pressure is necessary due to their extremely small hollow-core diameter compared with MLHCF [13,14]. Second, the fluidic sample can easily be guided into the side holes, which may affect the photonic band gap and induce tremendous loss. Therefore, to block the side holes in advance is a requisite step, which involves a complicated process. Third, the evanescent field and the cladding mode of HCPCF are not negligible, which gives rise to a strong background Raman signal of silica material. Then, a spatial filtering system with a small pinhole should be utilized to block the signal from cladding [11]. In addition, since the hollow-core size of PCF is very small, it requires a precise and intricate laser coupling system [11]. Last but not least, the numerical aperture of HCPCF is about 0.22, which means only 5% of the Raman radiation could be collected by HCPCF. These disadvantages make HCPCF not suitable for real-time monitoring.

MLHCF have better performance in terms of high sensitivity and fast online detection [7,14]. Compared with the wavelength dependence of photonic bandgap of HCPCF, the reflection of metal surface is almost wavelength independent and incident angle independent. MLHCF not only can support a large Raman shift range, but also have a large numerical aperture. Therefore, a larger portion of Raman radiation could be confined inside the hollow core [15,16]. Moreover, the penetration depth of metal surface is only hundreds of nanometers and the mode in hollow core cannot enter the outside silica cladding layer, which could greatly reduce the background fluorescence noise [17]. Most importantly, owing to the larger hollow-core diameter of MLHCF (0.3-1 mm), faster sample exchange can be realized within a few seconds even under normal pressure hence it is applicable for online detection [18]. MLHCF have been applied in measuring the Raman spectra of the liquids and gases in many researches [14–18].

With the goal of fast and sensitive online Raman detection, by using the method of volumetric enhancement of Raman scattering (VERS), we have designed a portable Raman cell of MLHCF to enlarge the light-sample interaction volume and improve the signal collection efficiency. The amount of Raman signal is nearly proportional to the length of MLHCF. To further enhance the light-sample coupling efficiency, a MLHCF inserting coupling method is proposed, which also makes our device to be easily packaged and robust for portable application. To enhance the signal collection efficiency, a reflector of gold coated fiber tip is introduced at the output, which not only reflects the forward transmitting Raman signal, but also the excitation light. In experimental demonstration, with a Raman cell of 3.1 cm in length, nearly 2 times enhancement of signal collection efficiency is observed by introducing a reflector and 49.3 times of enhancement in total is demonstrated compared with direct detection, which agrees well with the theoretical calculation. With a considerable flow rate under normal pressure, our system shows its potential application for fast online Raman detection, which also requires high-sensitivity and high-accuracy.

2. System structural design and experimental results

The schematic diagram of our configuration is shown in Fig. 1(a). The novelty of our setup lies in as follows. Firstly, we have proposed a fiber inserting method to enhance the light coupling efficiency between fibers, which can also be easily packaged to apply in portable detection. The optical coupling efficiency of inserting an optical fiber into the hollow-core fiber is considerably high as studied in our previous work [19,20]. Compared with optical coupling by the lens, our inserting method is more stable and easier to package. Secondly, for traditional reflection mode, only the reflected signal can be collected by Raman probe while transmission Raman light leaks out of SLHCF at the other side. To enhance the signal collection efficiency, a gold-coated glass fiber tip with the size just smaller than the hollow core diameter is introduced at output, which not only reflects the forward transmitting Raman signal, but also the excitation light. Theoretically, it provides 2 times of interaction length or interaction volume enhancement and 2 times of signal collection efficiency enhancement.

 

Fig. 1 (a) Schematic diagram of the system structure. The inset is the structure diagram of SLHCF; (b) The optical fiber is inserted into SLHCF and can guide both pump laser and Raman light. The sample is injected into SLHCF through the gap between the optical fiber and SLHCF; (c) The glass fiber coated with gold film is inserted into SLHCF, which can reflect both pump laser and Raman light. The sample leaks out of SLHCF through the gap between two fibers. Two inserting structures are sealed in three-way valves to ensure reliable sample exchange and fully light-sample interaction. (d) Experimental system of our configuration; (e) Details of coupling the Raman probe with the large-core optical fiber; (f) Gold film coated on the end face of the glass fiber.

Download Full Size | PDF

The experimental system is shown in Fig. 1(d). The large-core optical fiber (Ceramoptec WF275300, with diameter of 300 μm) for light coupling is inserted into one end of SLHCF (do-ko VSL320450) as shown in Fig. 1(b) and the glass fiber coated with gold film on the end face (Fig. 1(f), with thickness of 300 nm) is inserted into another end (Fig. 1(c)) for light reflecting, forming a folded Raman cell. Particularly, there is a tiny gap between the inner diameter of SLHCF (320 μm) and the outer diameter of the inserted fibers (300 μm), which could be used as the inlet and outlet for the fluid sample. Just only 10 μm gap could provide a significant flow rate even under normal pressure, yet its total area is equivalently to be a hollow core of 110 μm in diameter. Then we seal the inserting parts respectively in two three-way valves by ultraviolet glue to realize reliable Raman cell with continuous sample exchange through the branch of three-way valves. The optical path and sample flow path are overlapped along the SLHCF, providing a fully interaction between the pump laser and sample. The Raman probe (BWTEK) couples with the large-core optical fiber through a homemade coaxial ring, an adapter and a FC/APC connector as depicted in Fig. 1(e). It is worth noting here that the Raman probe not only can emit 785-nm laser for exciting, but also can collect Raman light for Raman spectrometer (BWTEK i-Raman Plus), while the inserted optical fiber acts as a bridge between probe and SLHCF. As a result, the pump laser can fully interact with the sample and the interaction volume is increased.

In order to demonstrate the volumetric enhancement mechanism in our structure, we keep all the experimental parameters same, for example, the power of 785-nm laser is 36.2 mW; the integration time of Raman spectrometer is 2 s; the length of SLHCF is 3.1 cm with inside sample volume of 2.49 μL, etc. At first test, we measure the Raman spectrum of pure ethanol by using a large core fiber tip only, which acts as an extended arm and can directly dip into the samples. The result is shown as the blue line in Fig. 2(a). Then we inject pure ethanol into SLHCF with a speed of 200 μL/min by syringe pump. Since the volume of the sample chamber is very small, we can calculate that the time needed to fill the capillary chamber or exchange the sample is less than 1 second. With the help of SLHCF to enlarge the interaction volume, the Raman spectrum is found to be enhanced by 28 times at 881 cm⁻¹ (C-C-O band), plotted as the red line in Fig. 2(a). Afterwards we insert the fiber tip with gold film into the other end of SLHCF and the Raman spectrum is obviously enhanced again, denoted as the black line in Fig. 2(a). With the reflector at the end of Raman cell, the signal intensity in the structure with gold film is increased by 1.76 times compared to that without gold film. Hence the total volumetric enhancement factor of Raman signal obtained from our structure is 49.3 times. Finally we flush the ethanol out with deionized water and the Raman spectrum is demonstrated as the pink line in Fig. 2(a), which proves that the sample has been completely pushed out of SLHCF and there is no residue in the capillary.

 

Fig. 2 (a) Raman spectra of pure ethanol in the beaker (blue line), SLHCF with (black line) and without gold film (red line). The inset is Raman peaks at 881 cm⁻¹ obtained by Gaussian fitting; (b) Raman peaks at 881 cm⁻¹ of pure ethanol in our configuration with different SLHCF lengths; (c) The relationship of normalized Raman intensity and SLHCF length.

Download Full Size | PDF

We can see that the major reason of the high enhancement factor in our configuration is due to the introduction of SLHCF, which provides a significant effect of volumetric enhancement of Raman scattering signal. In order to further demonstrate its volumetric enhancement, we continuously increase the length of SLHCF, and measure the Raman peak intensity of pure ethanol at 881 cm⁻¹ as shown in Fig. 2(b). We can see that the longer the SLHCF, the larger the volume of light-sample interaction and the stronger the Raman signal excited. Interestingly, we can also notice that the signal increasing is significant and linear at first, and gradually slows down after 3 cm as shown in Fig. 2(c). Later, we will analyze the VERS effect of SLHCF and give a full explanation. After considering the loss of SLHCF, we will find that the small the loss of SLHCF, the longer the linear region. We also will give a theoretical prediction of enhancement factor in our configuration.

3. Theoretical investigation

3.1 Volumetric enhancement of Raman scattering (VERS)

Generally, according to Eq. (1), the radiation of a particular molecule is related to its Raman scattering cross section of a molecule (σ_i) and the light intensity (I_i) exactly at its location. For the case of Raman detection for the sample with limited volume of V, the total power of Raman scattering is the summation of the radiation of all the molecules with number of M in the volume. If we assume that the molecules distribution and light intensity are uniform in V, then the total power excited in this volume could be $m\sigma V{I}_0$. It is proportional to the molecule density (m), the average light intensity of pump laser irradiation (I₀) and the volume of sample (V) [15].

Therefore, one way to enhance Raman signal is to increase the intensity of irradiation I₀ in this volume by using a high power pump laser. However it causes many problems, such as heating issues and sample damage [7]. Alternatively, we can increase the light intensity by focusing without increasing the pump power as shown in Fig. 3(a). For a tiny sample with volume of V (V is smaller than the focus volume), its Raman signal scattering is enhanced by a factor of 2 times when its experiencing light intensity is doubled by further focusing as shown in Fig. 3(b).

 

Fig. 3 The enhancement of Raman signal by further focusing for a tiny sample with volume of V.

Download Full Size | PDF

This method is highly suitable for the case with limited amount of analyte, where the pump laser beam could be focused to a tiny spot and be used to illuminate the tiny sample, even a single molecule [6]. Actually, the intensity of pump laser at the focus reaches the maximum and decays quickly to both sides. So the volume of laser-sample interaction is limited by the focus volume of the laser. Similarly, Cavity Enhanced Raman Spectroscopy (CERS) has achieved strong Raman signal by circulating the pump laser in the cavity to increase the effective interaction length or the light intensity of excitation [21]. On the other hand, SERS or TERS also originate from the localized field enhancement based on the near field effect of metal nanostructure. Only those molecules firmly attached right inside the tiny hot spot with dimension of a few hundreds of nanometers are excited intensely. Since the sample volume in the cavity or hot spot is small and the sample would firmly attached and cannot be rinsed off due to surface adhesion force, those methods are suitable for extremely small sample with limited amount of analyte and disposable samples.

On the other way, for the case with unlimited quantity of analyte, the situation is different. If we assume that a laser beam with total power of P_p passes through a big volume of sample in space, with a uniform power distribution at its cross section of S. Then the light intensity could be denoted as P_p/S, and the volume of sample illuminated could be expressed as SL (L is the length of light path). In this way, the total power of Raman scattering could be expressed as Eq. (2).

If the absorption rate of the sample is very low and the attenuation along the light path could be neglected, the total Raman signal scattering could be expressed as $m\sigma L{P}_{\text{p}}$. We can see that for fixed total power, the total radiation is unrelated with the light beam cross section S and it is proportional to the length of light path [6]. Therefore, we can elongate the light path to increase the signal intensity. Namely the focusing of beam does not contribute to the harvest of Raman signal in the space. Even through the light intensity is enhanced at focus point, the volume of sample illuminated is also reduced accordingly and linearly, with the total Raman signal keeping unchanged as shown in Fig. 4(a).

 

Fig. 4 Comparison of pump laser transmission in four structures. (a) Pump laser is focused by the objective and only the sample near the focus can be excited by pump laser to emit the Raman signal; (b) Pump laser is coupled into HCPCF. The laser at the focus is extended in HCPCF with low loss and the sample volume increases as the length of HCPCF increases; (c) Pump laser is coupled into MLHCF and transmits to a long distance; (d) Pump laser is coupled into MLHCF and reflected by the gold film at the other end of the fiber to double the volume of laser-sample interaction.

Download Full Size | PDF

Besides the total signal excited, we also need to consider signal collection efficiency. For the case of focusing, the intensive signal scattering from the focus point could be justly collected by the numerical aperture of objective lens. However, for the case of a light beam without focusing, the problem is that how to collect all the signal from such a large volume. The introducing of hollow core waveguide as shown in Figs. 4(b) and 4(c), is just for the volumetric signal collection mechanism, collecting most of the Raman radiation from the whole volume inside hollow core Raman cell and making the final signal collection efficiency as high as possible. Therefore, we can name this enhancement mechanism as volumetric enhancement of Raman scattering (VERS).

We can calculate the power and volumetric collection efficiency of all the Raman light excited in four structures as shown in Fig. 4 respectively, based on the assumptions that: (1) the intensity of pump laser is distributed homogeneously on the hollow-core cross section; (2) the transmission loss is negligible for both pump laser and Raman signal. As shown in Table 1, P₀ is the power of pump laser at the focus; S(x) is the area of the cross section of pump laser beam at position x; L is the length of Raman excited area or the length of hollow core fiber; D is the diameter of its hollow core; Ω_obj, Ω_HCPCF, Ω_MLHCF are the solid angles of the numerical aperture of the objective, HCPCF and MLHCF respectively. The expression of the power of excited Raman light is the same in the first three structures and only related to the length of light path or hollow core fiber, when the concentration of the sample is the same. In other words, the diameter of hollow core will not influence the total power of excited Raman light. Particularly for the case with a reflector at the other side as in Fig. 4(d), the total power of excited Raman light would be doubled, if we assume the reflectivity of gold film at inserted fiber tip is perfect.

Table 1. Comparison of all the excited Raman light and the volumetric collection efficiency of Raman light in four structures for the case of unconstrained quantity of analyte

View Table

Besides the total Raman excited, the signal collection efficiency is very important as well. For the case of Fig. 4(a), there is Raman signal excited all along the light path, however only the signal from the focus zone within the volume of S_focus·h could be collected by the lens (S_focus is the area of focused point; h is the depth of focus). Furthermore, the scattering signal is uniform distributed in all directions and only those parts of Ω_obj/4π within the numerical aperture could be collected. HCPCF and MLHCF provide better solutions, since they have large numerical aperture and can confine the Raman signal all along the fiber. At the output, their coupling efficiencies are Ω_obj/Ω_HCPCF and Ω_obj/Ω_MLHCF respectively, if the numerical aperture of objective is smaller than that of fibers. Furthermore, the volumetric collection efficiency of Raman light in Fig. 4(d) is also doubled, because the forward transmission Raman light can also be reflected and collected by the Raman probe. Therefore, the power of Raman light that can be collected in Fig. 4(d) is four times of that in Fig. 4(c) ideally.

3.2 Theoretical enhancement factor of SLHCF Raman cell with consideration of loss

If the transmission loss of fiber is neglected and the reflectivity of gold film is perfect, the introducing of SLHCF with a sufficient length would provide an infinite enhancement factor, while the introducing of the reflector can enhance another 4 times. But in real situation, with the consideration of losses, the effect would be weakened. We can calculate the enhancement factor of the total Raman signal intensity in our inserting Raman cell and compare it with the experimental results in Fig. 2(a). We can assume the scattering of excited Raman light is uniform in all directions. For the case of a bare fiber tip as shown in Fig. 5(a), the pump laser from the optical fiber tip diverges outwards at a solid angle of NA, and it only can collect the backward Raman light within the angle of 2α (2α is the total acceptance angle depending on the NA of the optical fiber). We assume that the distribution of light intensity at length l from the end face of the optical fiber is homogeneous and the power density of the pump laser is I₁. After an integral within its corresponding solid angle, the power of Raman signal collected by Raman probe can be approximated by Eq. (3).

 

Fig. 5 (a) Schematic of reflection Raman light in free space; (b) Schematic of reflection Raman light in SLHCF; (c) Schematic of the Raman cell in SLHCF. d₁ is the core diameter of the optical fiber (d₁ = 275 μm). d₂ is the outer diameter of the optical fiber or glass fiber (d₂ = 300 μm). D is the inner diameter of SLHCF (D = 320 μm).

Download Full Size | PDF

According to the Fresnel equation, the reflectivity is larger than 96% when the light with bandwidth of 800-1500 nm reflects on the silver surface at all angles. With the introducing of SLHCF, due to high reflectivity of silver layer, the pump laser and Raman scattering light in all directions will be confined in SLHCF and the intensity of Raman signal is related to the sample volume. As shown in Fig. 5(b), the optical fiber is inserted into SLHCF, which can guide pump laser and Raman light at the same time. We assume that the distribution of light intensity on the cross section of SLHCF is homogeneous. The total power of backward Raman signal P₀ collected by Raman probe can be approximated by Eq. (4) [15].

 

Fig. 6 (a) The enhancement factor γ₁ of the Raman light in SLHCF which can be collected by Raman probe versus that in free space at different fiber lengths; (b) The relationship between normalized reflection and transmission Raman intensity and SLHCF length. P₂, P₃ is the reflection Raman light in the length of 0-L and L-2L and P₄, P₅ is the transmission Raman light in the length of 0-L and L-2L separately. P_total is the sum of P₂, P₃, P₄ and P₅; (c) The enhancement factor γ₂ of the Raman light in SLHCF with gold film versus that without gold film at different fiber lengths; (d) The enhancement factor γ (γ = γ₁ × γ₂) of the Raman light in SLHCF with gold film versus that in free space at different fiber lengths. In our experiment, the length of SLHCF is 3.1 cm, which can reach 60.4 times of enhancement factor in theory.

Download Full Size | PDF

The schematic of our configuration is demonstrated in Fig. 5(c). A glass fiber coated with gold film is introduced to further enhance the signal collection efficiency. Due to the reflection of pump laser, the effective laser-sample interaction length or the total pump laser is doubled compared to the structure without gold film. Compared with double-length SLHCF, the returning length of Raman light to Raman probe is shorter and the transmission loss is smaller. Furthermore, gold film not only can reflect pump laser, but also can reflect the forward transmitting Raman light, which can also contribute to the enhancement factor. As a result, the total Raman signal is composed of four parts: the backward and reflected forward Raman light excited by the pump laser transmitting from the optical fiber to the gold film; the forward and reflected backward Raman light when the pump laser transmits from the gold film to the optical fiber, which can be approximated by P₂, P₃, P₄, P₅ in sequence.

We can calculate the total volumetric enhancement factor γ ($\gamma ={\gamma }_1\times {\gamma }_2$) as shown in Fig. 6(d). γ increases with SLHCF length of 0-3 cm. The longer the fiber length, the larger the sample volume. Hence the pump laser can interact with more sample and more Raman light can be excited. Light transmission loss is higher with longer length. When the length is around 3 cm, the loss and volumetric enhancement achieve balance and thus γ reaches the maximum. After that, γ decreases with longer length. However, the curve change trend does not agree well with that in Fig. 2(c) because the gold film is defective (Fig. 1(f)), which will reduce its reflectivity and the reflection area. In theory, Raman signal can be enhanced by 60.4 times in our Raman cell compared to that with a bare fiber tip. The enhancement factor is 49.3 times in our experiment, which agrees well with the theoretical result.

4. Demonstration of application in sensitive and fast Raman detection

4.1 Raman spectra of ethanol-methanol mixtures

For the purpose of verifying the detection accuracy and reusability of our device, we try to detect the mixture of ethanol and methanol. First of all, we detect the Raman spectrum of pure ethanol (Fig. 7(a)) in our configuration. Then we exchange the sample and rinse pure ethanol with pure methanol. When the ethanol sample is just fulfilled the chamber, the Raman spectrum is shown as Fig. 7(b). We can see that Raman peaks at 881 cm⁻¹ and 1277 cm⁻¹ of pure ethanol have completely disappeared in the Raman spectrum of pure methanol, which verifies that the ethanol in SLHCF has been completely displaced by the methanol and there is no ethanol residue in the capillary. Finally, we mix two liquids at the volume ratio of 1:1 and fill the SLHCF with the mixture. As shown in Fig. 7, Raman peaks in the mixture correspond to those in the pure liquid. According to the Raman signal intensity at 881 cm⁻¹ and 1039 cm⁻¹ (C-O stretch), we can calculate the concentration of ethanol (48.4%) and methanol (51.6%) in the mixture respectively. These results indicate that our device could be used to estimate the concentration of the various compositions in the mixture. Most importantly, since there is no signal residual after sample exchange, our device is very good at fast online detection.

 

Fig. 7 Raman spectrum of (a) pure ethanol, (b) pure methanol and (c) the mixture of ethanol and methanol at the volume ratio of 1:1.

Download Full Size | PDF

4.2 Raman cell for monitoring antibiotics

In addition, we have applied our device in detecting antibiotics. CTX (cefotaxime sodium, CAS: 64485-93-4) is the third-generation cephalosporin and a parenteral broad-spectrum antibiotic, which has been widely used for chemotherapy of serious infections [22]. Monitoring the concentration of the antibiotics is important to prevent antibiotic abuse. The techniques for detecting cephalosporin antibiotics include high performance liquid chromatography, immunoassay, microbial analysis, fluorescence analysis, etc [23,24]. These methods have the disadvantages of expensive equipments, complicated sample preparation, long analysis time and not suitable for on-line detection. Therefore, it is of great significance to develop a rapid and accurate method for detecting antibiotics.

In our experiment, we dissolve CTX powder in water with different concentration varying from 0.25 to 8 mg/mL, which is within the range of current medical use [25]. The laser power and integration time are adjusted to 90.8 mW and 20 s respectively due to weak Raman signal. We remove the background noise of the source data by adaptive iteratively reweighted penalized least squares (airPLS) model and smooth the spectra by base-line wavelet model [26,27] as shown in Figs. 8(a)-8(f). Raman spectra of CTX solution exhibits prominent Raman peaks within 1400-1700 cm⁻¹ [28] and we pick the characteristic peak at 1587 cm⁻¹ (C = C band, thiazine ring) to quantitatively measure the concentration of CTX in the solution. As demonstrated in Fig. 8(g), Raman peak intensity is linearly dependent on the concentration of CTX. The result indicates that our Raman cell can detect the concentration of antibiotics with high sensitivity in a broad range, hence has a promising future by means of both qualitatively and quantitatively monitoring bio-analytes.

 

Fig. 8 (a)-(f) Raman spectra of CTX with different concentrations ranging from 0.25 to 8 mg/mL; (g) The linear relationship between the Raman signal intensity and the concentration of CTX.

Download Full Size | PDF

5. Conclusion

In this paper, we demonstrate a Raman sensor based on SLHCF and VERS with high stability and reusability. The silver and gold film reflect most pump laser and Raman light to increase the laser-sample interaction volume and enhance the collection efficiency of Raman light. In the experiment, Raman light intensity of our configuration is increased by 49.3 times compared with bare fiber tip, which agrees well with that in theory (60.4 times). Higher enhancement factor could be achieved if we further improve the production process of the silver coating or replace it by the gold coating to reduce the fiber loss. Oxidation of the silver coating is not obvious because the power of pump laser required to detect liquid is relatively lower. In addition, Raman signal intensity of the sample varies linearly with the concentration, which indicates that our Raman cell can realize highly sensitive and accurate detection of the sample concentration. Only micro-liter sample is needed and exchanged very quickly in SLHCF. Tiny cross contamination when exchanging samples does not make sense. Therefore, our Raman cell can be a flow structure for real-time monitoring and all the samples go through the inspection in a polling mode. Our well-designed coupling window and well-encapsulated structure make it more portable and reduce the background noise from the glass wall. These merits fully prove that our structure has a promising future in online Raman bio-sensors.