1. INTRODUCTION

Due to its rotational–vibrational absorption lines, acetylene gas is a useful optical reference between 1510 and 1540 nm [1] used in telecommunications. As a result, pressure-broadened acetylene vapor cells such as the National Institute of Standard Technology (NIST) Standard Reference Material (SRM) 2517a are used in Swann et al. [2] to calibrate tunable laser frequencies, optical spectrum analyzers, and other light sources in the near infrared region. The fiber-coupled glass cell at acetylene pressures of approximately 50 torr gives 15 lines with uncertainties of approximately 10 MHz. The NIST reference cells play an important role in calibration, but have a relatively large footprint due to free-space optical components. As lasers and components are made smaller, gas-filled hollow-core fibers offer an alternative to current technology.

With the advent of low-loss hollow-core photonic bandgap fibers (PBGFs) [3,4], studies were started on Raman scattering [5,6], generation of megawatt optical solitons [7], gas sensing [8], and electromagnetically induced transparency [9,10] using PBGF. Due to its portability and the significant interaction of light and gas inside PBGF, it was advantageous to use this optical device as a portable frequency reference [11]. For that purpose, various techniques have been developed to create photonic microcells (PMCs), which vary depending on the applications.

A variety of techniques have been developed to make sealed PMCs. For example, Triches et al. [12] generated a 2.7 m, 0.1 Torr optical reference using the encapsulation technique to seal the hollow-core fiber. In this technique, free-space optics are used to couple the light into the PMC. Light et al. [13] made a long PMC by splicing both ends of a hollow-core photonic crystal fiber (HCPCF) to conventional fibers. The 1 m long HCPCF was initially filled with acetylene and helium ranging from 3750 Torr (5 bars) to 7500 Torr (10 bars). Then, helium diffuses throughout the hollow-core fiber. Furthermore, Light et al. [14] built an optical fiber frequency standard using a collapsing technique with relatively low transmission efficiency due to the distortion structure of the fiber in the collapsing region. Wang et al. [15] also created long sealed PMCs based on the latter work inside a 1 m acetylene-filled PBGF and characterized the accuracy of the PMCs. The aforementioned fabrication techniques are appropriate for long PMCs, and when low coupling efficiency is not a limitation.

For short PMCs, a new technique is needed. In this paper, our goal is to produce a compact reference made with a novel (to our knowledge) fabrication technique with lengths and pressures similar to the NIST SRM’s for moderate-accuracy frequency measurements, as described in Luder et al. [16]. We will introduce a short, portable, all-fiber compact reference that has comparable accuracy to NIST SRM 2517a on the line center to calibrate optical devices without the need for free-space coupling. This device can be fit into a laser, can be mounted onto photodetectors (PDs), and can be readily connected to large-core multimode fibers (MMFs) such as 400 μm core MMF when connectorized.

Here, we demonstrate two different methods of fabricating short PMCs called “Q-tipping” and “tapering.” Thereafter, characterization of these PMCs, in terms of transmission efficiency and other parameters, is described. One advantage of the tapering technique is the reduction of an etalon-like effect, resulting in satisfactory cell performance for many applications. PMC partial gas pressures were calculated from absorption fitting using results from [17] and [18] and HITRAN parameters [19]. Furthermore, the accuracy of the PMC is checked with the saturated absorption spectroscopy setup described in Knabe et al. [20] and Eq. (1) of Thapa et al. [21]. Although connectorization of standard optical fiber is a well-documented procedure [22], our work [16,23] introduces it to the construction of short PMC references. Since unconnectorized PMCs exhibit a large numeric aperture [16], connectorization allows two fibers to have physical contact, so light can be coupled directly into MMFs. Additionally, a patent has been filed for making short PMCs [24]. Finally, the long-term stability of a few cells has been measured for the two different methods of fabrication of the PMCs.

2. CONSTRUCTION OF THE PMC

Construction of the PMC includes three steps: (1) splicing a 20 μm core PBGF (HC19-1550) to a telecommunication-grade single-mode fiber (SMF) (Corning SMF28E), as shown in Fig. 1(a), (2) placing the open end of the PBGF in a vacuum setup, evacuating and filling the fiber with acetylene gas, and (3) tapering or Q-tipping the fiber to seal the PMC. As shown in Figs. 1(b) and 1(c) on the left side, a cone-shaped pattern called a “needle” is created during the process of Q-tipping or tapering the core and cladding. Figure 1(b) shows a Q-tipped PMC, and Fig. 1(c) shows a tapered PMC.

 

Fig. 1. (a) Camera image of the splice between a 20 μm core PBGF (HC19-1550 Thorlabs) on the left-hand side and SMF (SMF28E) on the right-hand side, (b) camera image showing the Q-tipped end of the PBGF, (c) camera image showing the tapered end of the PBGF.

Download Full Size | PDF

A. Splicing the PBGF to SMF

The first step toward making a PMC is splicing an approximately 1 m length of PBGF to a SMF. There are different splicers with modified recipes to splice the PBGF to SMF, as shown in [25], in which an arc fusion splicer is used. Here a Vytran filament fusion splicing system [FFS-2000] is used for this step, and the recipe is modified for this specific machine. The PBGF is commercially available and is chosen such that we can achieve appreciable mode matching to other fibers.

The splicer is calibrated with a normalization routine, and then an optimized splicing program recipe is applied to splice the SMF and PBGF interface, giving us a half-cell. Figure 1(a) shows an image of the splice taken using the splicer camera. A protection sleeve is melted on the splice of SMF to PBGF to protect the splice. The quality of the splice is tested by measuring its transmission. We have been able to achieve typical transmission of approximately 90%, and up to a maximum of 95%, using the modified splice program.

B. Evacuating the Spliced Fibers and Filling with Acetylene

The open end of the half-cell is attached to a vacuum chamber with a custom-made feed-through and is evacuated to mTorr levels for approximately 24 hours using both roughing and turbo pumps. The chamber is evacuated for a day, and the setup is filled with acetylene at a given pressure for an hour. In our case, the acetylene pressure in the vacuum chamber varies from 7.3 to 80 Torr. P13 or P11 absorption lines of acetylene are monitored while the system reaches equilibrium, which is assessed by the stability of the absorption line. The vacuum chamber is attached to a capacitance manometer [MKS instruments XDCR500T2-¾-CFROT 0.25%] that gives the pressure readout. PMCs are made for different combinations of length and pressure.

C. Tapering or Q-Tipping the Open End of PBGF

The final step in making the PMC is sealing the open end of the PBGF. This can be done using two different methods: tapering or Q-tipping. The end of the PBGF can be Q-tipped while acetylene gas is trapped inside of it. Specifically, about one inch of the PBGF fiber is stripped using a razor blade and inserted into the arc fusion splicer. A brief pulse of current causes the PBGF fiber to collapse, and a ball of glass called a “Q-tip” forms, sealing acetylene inside the PMC, as shown in Fig. 1(b). Alternatively, the PBGF can be sealed by the “tapering” method. During this process, the splicer simultaneously pulls and heats the PBGF into an hourglass shape that finally separates the PBGF, as shown in Fig. 1(c). We use the Ericsson fusion splicer (FSU 995FA) with modified settings for these processes.

The recipe for Q-tipping the PBGF is customized by adapting a program, “normal SM + SM,” that splices two SMFs, as shown in Table 1. In this program fusion currents are optimized such that a Q-tip forms on the end of the PBGF. The optimum parameters produce the least distorted, most symmetric shape of the Q-tip. While most parameters can stay the same, fusion current 1 needed to be adjusted by about 2 mA when the electrodes were replaced.

Table 1. Recipe for Q-Tipping the Open End of PBGF

View Table | View all tables in this article

The recipe for tapering the fiber adapts the program “SM fiber lens,” as shown in Table 2. This program tapers an SMF using a pulling process. In addition, it is optimized to taper the PBGF. To apply the process of tapering to PBGF, just the first round of pulling (Pull 1) is used. Furthermore, fusion time 1 is user determined and is manually stopped when the process of sealing the PBGF is done. The process of tapering PBGF takes about 5–8 s, and it depends on the pressure to which the PBGF is filled with acetylene.

Table 2. Recipe for Tapering the Open End of PBGF

View Table | View all tables in this article

These optimum parameters were found by first optimizing the tapering and separation of a single air-filled PBGF fiber into two parts, and then testing it on acetylene-filled fibers. At 13 mA, the PBGF did not reliably separate into two tapered fiber ends, but at 15 mA, the tapering program performed reliably. The program was optimized to create the geometry shown in Fig. 1(c), with about 300 μm extending beyond the needle. This gives enough length to allow connectorization, including cleaving and polishing, without any gas leakage, and is short enough to prevent reduced coupling efficiency. Once it is connectorized, the length of glass extending beyond the needle is about 150 μm.

3. CHARACTERIZATION OF THE PMC

The specifications of the PMC are given in terms of its transmission, absorption, coupling loss to MMFs, accuracy in finding the line center, and total pressure.

A. Transmission, Absorption, and Coupling Measurements

One challenge in creating a PMC is achieving high off-resonant optical transmission while it contains gas. The off-resonant transmission of the PMC is given by the ratio of the power incident on the SMF to the power output at the tapered or Q-tipped end of the PMC.

In order to measure absorption, the laser frequency was swept across the P13 absorption line of acetylene and the ratio of the transmitted power on-resonance to the power off-resonance was measured at the output of the PMC. The best tapered PMC transmission was 82%, with absorption of 50%, for which the PMC length was 10 cm and the vacuum setup was filled to an acetylene pressure of 36 Torr. Many cells were created with 75%–80% transmission.

One reason to use the tapering method was that the Q-tipping method results in a lower off-resonant transmission compared to the tapering method. The best off-resonant transmission achieved by Q-tipped cells was 37%, which is a factor of 2 smaller than the best tapered cell off-resonant transmission.

Coupling efficiency between the PMC and various MMFs was determined by measuring the transmission through butt-coupled 400 μm, 200 μm, and 62.5 μm core MMFs. The experimental setup shown in Fig. 2 is used to measure this parameter where the unconnectorized PMC is butt-coupled with MMF by careful manual alignment in the Ericsson fusion splicer. The light from a tunable fiber-coupled diode laser near 1532 nm is coupled into the SMF part of the PMC, which has a fiber-optic connector/ angled physical contact (FC/APC) connector. When the cell contains acetylene, care is taken to tune the laser off-resonance to avoid significant cell absorption. As shown in Fig. 2, the power measured out of the SMF patch cable is recorded as power level A, and can be remeasured throughout the process. A large-area PD is brought as close as possible to the tapered end to record the transmitted power B, and tapered cell transmission is B/A. Finally, light is coupled from the cell into an MMF, and the transmitted power is measured at C, making the insertion loss of the whole device equal to C/A.

 

Fig. 2. Power measurements used to characterize cell transmission efficiency. A, power measured out of fiber-coupled laser source; B, power measured after collapsing fiber cell; C, power measured through 400 μm [BFL48-400 THORLABS], 200 μm [FT200EMT THORLABS], and 62.5 μm [GIF625 THORLABS] core multimode fibers. All power measurements are made off-resonance in the case of acetylene-filled cells. The laser wavelength is approximately 1532 nm.

Download Full Size | PDF

Table 3 shows the optimized coupling transmission through various 400 μm, 200 μm, and 62.5 μm core MMFs. In addition, Table 3 gives the measured fractional transmission through two different tapered PMCs. We report results for PMCs no. 52 and 53 with lengths of 6 cm and fractional absorptions of 25%.

Table 3. Transmission and Insertion Loss for 20 μm Core Tapered Unconnectorized Cell Nos. 52 and 53

View Table | View all tables in this article

While transmission efficiency is high, the mode of the beam exiting the collapsed fiber is highly divergent, and therefore not readily coupled into SMF. However, coupling into MMF was demonstrated, with efficiency increasing with core size. These values therefore give an indication of the mode shape and are informative for applications requiring fiber transmission of the output from the PMC.

B. Pressure Broadening and Acetylene Pressure Measurements

To characterize the gas purity and pressure, we examine the pressure broadening and integrated optical depth and deduce the partial pressures of acetylene for each PMC.

1. Setup Schematic

To carefully measure absorption of the cells, a tunable laser is coupled into the PMCs and the frequency is scanned across the feature, as shown in Fig. 3. Next, the laser piezoelectric rransducer (PZT) ramp voltage and the transmission through the fiber ring cavity and the PMC were measured simultaneously as a function of time using two Red Pitayas for data acquisition (DAQ). Ring cavity output, used for frequency calibration, is integrated with the same beam of light to the large-area detector (IR photoreceiver, New Focus). The connectorized PMC is connected to the large-area detector (IR photoreceiver, New focus). PDs are connected to DAQ systems, and the signal generator is directly connected to DAQ system 1, to which PD 2 is also connected. Both DAQ systems are synced to prevent signal delay.

 

Fig. 3. Schematic for characterizing the pressure inside PMCs: a diode laser [SANTEC tunable semiconductor laser TSL-210] works in the 1.5 μm range. Ring cavity is for relative frequency calibration. PD 1 and PD 2, large-area detectors [IR photoreceiver] for inspecting the ring cavity transmission and PMC P13 absorption line; FC, 30%–70% fiber coupled beam splitter; OI, optical isolator. Two Red Pitayas are used as DAQ systems. Signal generator and ring cavity transmission to PD 2 are connected to DAQ 1 channels 1 and 2. PMC absorption to PD 1 is connected to DAQ 2 channel 1. Blue dashed lines are electrical signals.

Download Full Size | PDF

Figure 4(a) shows the calibrated data, after it was taken by the 14-bit Red Pitaya interface used as a DAQ system. This data set shows tapered cell no. 48’s transmittance to PD 1. Figure 4(b) shows the ring cavity transmission to PD 2, and the ramp voltage is taken by DAQ 1 channel 1. In Fig. 4(b) the ramp voltage sweeps from low to high laser frequencies. For both Figs. 4(a) and 4(b), the $x$ axis is calibrated by using the ring cavity, and then the frequency is calculated with respect to the P13 line frequency. In order to find the pressure broadening, the optical depth is fitted with a Voigt profile. To attain the frequency-dependant optical depth $\alpha (\nu )L$ from the measured absorption line, Beer’s law $I=I_0{e}^{-\alpha (\nu )L}$ is used, where $I$ is transmitted intensity, $I_0$ is incident intensity, $L$ is optical path length, and $\alpha (\nu )$ is the absorption coefficient.

 

Fig. 4. (a) Calibrated transmittance for PMC no. 48 registered by PD 1 in Fig. 3. The feature on the left side of the P13 line shows the ${\mathrm{H}}^{12}{\mathrm{C}}^{13}\mathrm{CH}$ absorption line (second isotopologue of acetylene) approximately 1500 MHz away from the P13 peak. (b) Calibrated data for ramp voltage registered by DAQ 1 channel 1, which is used to drive the PZT of the tunable laser and ring cavity transmission derived from PD 2 with free spectral range $(\mathrm{FSR})=93\pm 1\mathrm{MHz}$. The green line shows the ring cavity transmission, and the red line is the ramp voltage.

Download Full Size | PDF

Analysis is done in Python using a fitting module called lmfit [26]. The optical depth is expected to follow a Voigt profile, which is a convolution of a Gaussian, due to Doppler broadening, and a Lorentzian, due to pressure broadening [17]. The Gaussian full width at half-maximum (FWHM) is $2\sigma \sqrt{2\ln 2}$ [18], and $\mathrm{\Gamma }$ is the Lorentzian half of the FWHM in the fitting code. To fit the data to a Voigt profile, $\sigma$ is fixed to 201.4 MHz in the code for the case of acetylene gas at room temperature [18].

Once the Gaussian FWHM is fixed, then the optical depth is fitted to a Voigt profile and a best-fit value is obtained for $\mathrm{\Gamma }$. To extract the pressure from the measured pressure broadening for the case of a pure cell that is filled only with acetylene, the Lorentzian FWHM ($2\mathrm{\Gamma }$) will be found in the first step. Then the pressure will be found by dividing the Lorentzian FWHM by the pressure broadening factor for the P13 line extracted from Swann et al. [2]. According to NIST’s extensive studies on acetylene broadening in these pressure regions, the collisional or pressure broadening parameter is 11.4 MHz/Torr for the P13 line [2]. The process of finding the pressure can be done for different acetylene lines using the pressure broadening parameter and Lorentzian FWHM of that particular line. Figure 5 shows the fitted optical depth to a Voigt profile for PMC no. 53. The difference between the optical depth and the fitting of PMC no. 53 is shown as residuals in Fig. 5.

 

Fig. 5. PMC no. 53 fitted with a Voigt profile to find the pressure broadening with residuals.

Download Full Size | PDF

2. Measuring Partial Pressures

From the measured $\alpha (\nu )$, the partial pressures of acetylene and air contaminant are determined. First, the acetylene partial pressure (i.e., pressure of absorber) is calculated by integrating $\alpha (\nu )L$. Next, the air contaminant pressure is found from the pressure broadened width, plotted as $\alpha (\nu )L$ versus frequency. Here, “air” is a gas that contributes to the total pressure but not to the absorption.

The HITRAN data for the P13 acetylene line strength of the HITRAN parameter [19] and the relationship between the optical depth and the pressure of the ideal gas due to acetylene are used to find the acetylene pressure. Then the frequency-dependent absorption coefficient is given by

Having the acetylene partial pressure $P_a$ from Eq. (3) and Lorentzian FWHM $2\mathrm{\Gamma }$ determined from the Voigt fit, the “air” partial pressure can be determined using the equation $2\mathrm{\Gamma }=2{\gamma }_c{P}_c+2{\gamma }_a{P}_a$, where ${\gamma }_c$ is the contaminant broadening factor, $P_c$ is the contaminant partial pressure, and ${\gamma }_a$ is the acetylene broadening factor. HITRAN values for broadening factors are $2{\gamma }_a=\mathrm{11.40.6}\mathrm{MHz}/\mathrm{Torr}$ and $2{\gamma }_c=6.14\mathrm{MHz}/\mathrm{Torr}$ (considering the contaminant in our case to be air) [19].

3. Result for PMC No. 53 and Half-Cell A

Through the Voigt fitting of Fig. 5 and analysis of Section 3.B.2, the Lorentzian HWHM Γ, pressure of absorber $P_a$, and contaminant pressure $P_c$ were calculated for connectorized tapered cell no. 53, as shown in Table 4. Before sealing PMC no. 53, the vacuum setup was filled to 12 Torr with acetylene. The same analysis was done as a comparison for a half-cell, which is a PBGF with one end spliced to SMF and the open end attached to a vacuum chamber with a custom-made feed-through. The half-cell was filled to the chosen acetylene pressure, and the following parameters were found for half-cell A, when the vacuum setup was filled to 14 Torr.

Table 4. Characterization of PMC No. 53 and Half-Cell A

View Table | View all tables in this article

4. Comparing Measured Absorption for a PBGF Half-Cell and a PMC to SpectralCalc

To verify the partial pressures found above, the frequency-dependent optical depth was computed using SpectralCalc for values from Table 4 for PMC no. 53 and half-cell A. SpectralCalc is a software package [27] that computes the optical depth of the gas cells for a given set of length, temperature, and partial pressures based on assumed line shapes and HITRAN data. Assuming no level of contamination, the PMC absorption computed using SpectralCalc was not in agreement with the measurement. Better agreement was observed by computing absorption with a percentage of air contamination as determined from the fitting. In contrast, partial pressure measurements done for half-cell A show that no air contamination was introduced to the setup when the half-cell was held in the vacuum setup filled with acetylene.

C. Contamination Characterization

To understand contamination, the partial pressures were measured in half-cell B, and subsequently two PMCs were made from it by Q-tipping and re-Q-tipping it. Half-cell B was uncontaminated, while the Q-tipped PMCs show some contamination, as in Table 1 of [23]. When PMC no. 58 was made from half-cell B, the acetylene partial pressure decreased from 9.4 to 7.7 Torr. Similarly, when PMC no. 59 was made from PMC 58 by re-Q-tipping, the pressure decreased to 5.7 Torr while the contamination pressure increased. We verified that due to some unknown physical process, contamination increases each time the PMC is collapsed, while the acetylene partial pressure simultaneously decreases. Contamination happens for the tapered cells as well. The source of the contamination was not conclusively identified. We initially speculated that it might be due to gas diffusion through hot glass, but those rates appear to be much too slow, based on the diffusion coefficient of oxygen through the fused silica derived from Fig. 5 of Williams [28]. Other processes responsible could be leakage or outgassing during tapering and Q-tipping, and further work will be required to reduce this contamination if needed. Generally, the contamination does not significantly affect the application of the PMC as a moderate accuracy-frequency standard device.

D. Accuracy

The center wavelength of the pressure-broadened feature for connectorized, tapered PMC no. 53 was measured to compare with the description of NIST SRM 2517a [1]. The setup shown in Fig. 3 was integrated with the sub-Doppler gas-filled HCPCF described in [20]. The center of the sub-Doppler feature was found to be 10 kHz [20]. Then the PMC optical depth was recorded simultaneously with the sub-Doppler feature for six sets of data for the same PMC no. 53. The sub-Doppler reference (fiber length 4.5 m) was filled with acetylene to 170 mTorr, while cell pressure was on the order of Torr. The sub-Doppler feature was shifted from the center of the Doppler because the probe was shifted by 55 MHz from the pump using an acousto-optical modulator to decrease the noise due to pump and probe interference. The sub-Doppler optical depth was fitted to Eq. (2) of Ref. [21], while the PMC was fitted to a Voigt profile. The frequency separation between the sub-Doppler and PMC varied by less than 10 MHz for six sets of data. Figure 6 shows the experimental data taken for the small shift between the sub-Doppler and cell center. The maximum value of the shift separation between the sub-Doppler and cell center for six sets of data for the same cell was $15.4\pm 3.3\mathrm{MHz}$. Although the cell is highly contaminated, the experimental data show that this separation is small for the PMC P13 line center located at approximately 195 THz.

 

Fig. 6. Comparison of saturated absorption spectroscopy (SAS) reference simultaneously with optical depth of connectorized PMC no. 53 and SAS residuals. $X_{\mathrm{Cell}}$ is the center of the PMC P13 line, $X_{\mathrm{SD}}$ is the center of the sub-Doppler, and $X_D$ is the center of the Doppler. The solid red lines show the fittings to the sub-Doppler and cell centers.

Download Full Size | PDF

 

Fig. 7. (a) Picture of the connectorized tapered PMC no. 53, (b) schematic of a connectorized tapered PMC.

Download Full Size | PDF

E. Etalon-like Effects

Errors in the line center are due to multipath interference observed as an etalon-like effect. Due to this etalon-like effect, Q-tipped cells show amplitude and frequency modulations on the wings of the absorption line. This modulation is more greatly reduced for tapered cells than for the Q-tipped cells, as shown in Fig. 2 of [23]. The Q-tipped cell exhibits an etalon-like modulation in the transmitted intensity with two components. The largest amplitude of 13% has a frequency of 2 GHz, while a smaller amplitude of approximately 2% is observed with a 25 MHz frequency.

We were able to eliminate the 2 GHz modulation by using the tapering technique, but the 25 MHz modulation is still present in both Q-tipped and tapered cells. The characteristic length for the larger frequency modulation is approximately $L=5\mathrm{cm}$, which is estimated using $f=c/2\mathrm{nL}$, where $c$ is the speed of light in a vacuum and $n$ is the fused silica group index at 1550 nm. Since the aforementioned characteristic length is close to the length of the Q-tipped PBGF in the PMC, this modulation may arise due to the backreflection of the light from the Q-tipped end and the PBGF to the SMF spliced end. In contrast, the characteristic length of the modulation with a small frequency is approximately 4 m and is relatively small in amplitude.

The etalon-like effect on the wing also masks the ${\mathrm{H}}^{12}{\mathrm{C}}^{13}\mathrm{CH}$ line close to the P13 line. ${\mathrm{H}}^{12}{\mathrm{C}}^{13}\mathrm{CH}$ has a peak approximately 1500 MHz away from the P13 line center, which can be used as a frequency calibration check. As shown in Fig. 2 of Hosseini-Zavareh et al. [23], the ${\mathrm{H}}^{12}{\mathrm{C}}^{13}\mathrm{CH}$ line is masked by an etalon-like effect.

4. CONNECTORIZED CELLS

Because light exits the fiber in a cone-shaped spatial pattern, coupling into or splicing to an SMF after collapse has proven lossy, as described in [8]. Our solution is to connectorize the bare fiber ends so the PMC light can be coupled directly onto PDs and into MMFs.

A. Connectorizing Process

Two different processes were developed for connectorizing PMCs. For the first method (called the “tapering method”), the fiber ends were tapered, placed in a connector, and polished, as is similarly done with any type of step-index fiber. First, the epoxy (Thorlabs F112, 24 hr cure time) is injected into the ferrule. Second, the PMC tip is fitted into a 126 μm standard FC/PC connector (FIS F12070900). Third, the cell tip is scribed roughly 150 μm away from the needle of the tapered cell, as shown in Fig. 1(c), while being viewed under the microscope. When the epoxy is cured, the cell tip is polished. Since the width of the tapered cell is approximately 100 μm, it fits into standard fiber-optic connector/ flat physical contact (FC/PC) connectors. Standard connectorization of the very short tapered cells is easier, making them more robust inside the connector. With that said, we were able to make a connectorized tapered cell with 12% transmission to the large-area detector with an absorption of 25%.

Q-tipped PMCs were also polished and connectorized by a method called the “Q-tipped method.” For the Q-tipped PMC, the tip is inserted into a custom-drilled connector (FIS F12772190) that extends 300 μm from the ferrule. Then the epoxy (Thorlabs F112, 24 hr cure time) is injected into the connector and fills the connector. The epoxy does not cover the tip of the Q-tipped cell and ferrule. Next, the tip of the PMC is pulled back until the ferrule end and tip are even. After the epoxy is cured, the ferrule is cleaned and no polishing is needed. The best off-resonant transmission achieved using Q-tipped connectorization was 4%. Figure 7(a) shows the picture of the final product and Fig. 7(b) shows the schematic of a connectorized tapered PMC.

B. Fitting on Connectorized Cell No. 53

PMC trial no. 53 is the prototype of a tapered cell that survived after connectorization. Optical fitting was done before and after connectorization on the cell to check if there was any gas leak after polishing. Measured and remeasured partial pressures matched within the error bars, and no outgassing was measured even after connectorization.

5. LONG-TERM STABILITY OF THE CELLS

One important question for sealed fiber cells is how stable they are in terms of partial pressures over time. We addressed this question by remeasuring the partial pressures for tapered connectorized PMC no. 53, Q-tipped unconnectorized PMC no. 17, and Q-tipped unconnectorized PMC no. 19. The PMCs were fabricated via the two different techniques of tapering and Q-tipping, and they had been previously measured. While measurements on PMC nos. 17 and 19 are separated by one year, measurements for cell no. 53 are separated by approximately 6 months.

The results shown in Table 5 indicate no change in acetylene partial pressure within the error bars, and are certainly inconsistent with a leak. These errors are associated with the extracted off-resonant transmission for absorption data shown in Fig. 4(a), which causes error in finding the partial pressures. The key results are tabulated in Table 5. In addition, the line center shift doesn’t change because, as Table 5 shows, the partial pressures remain constant within error bars.

Table 5. Long-Term Stability of Connectorized Tapered PMC No. 53 and Unconnectorized Q-Tipped PMC Nos. 17 and 19

View Table | View all tables in this article

6. CONCLUSION

Short acetylene-filled PMCs (6–10 cm) were fabricated in the pressure-broadened regime, in the form of a connectorized all-fiber device that can be mounted into PDs for moderate-accuracy frequency measurements and may be connected to 400 μm core MMF. These PMCs can fit into laser devices as an optical reference. In this work, the butt-coupling transmission increased to 61% into a 400 μm core MMF and 52% into a 200 μm core MMF, and the 13% etalon-like effect that appeared on the wings of the optical depth in the Q-tipping method was reduced to less than 1% by using the tapering method. We were able to find the line center shift of the tapered connectorized PMC no. 53 with an accuracy of 10 MHz, which is comparable to the NIST SRM 2517a accuracy on the P13 absorption line center. Moreover, experimental data for the frequency between a sub-Doppler reference and the connectorized tapered PMC no. 53 center shows $15.4\pm 3.3\mathrm{MHz}$ separation. Although some PMCs are contaminated between 50% and $-86\%$, contaminant pressure only causes a small shift of less than 15.4 MHz on the PMC center with respect to the sub-Doppler feature, and the PMCs are still useful devices for moderate-accuracy frequency measurements. Finally, results of the long-term stability show no changes in total pressure within uncertainties over a period of at least 5 months for the PMCs.