1. Introduction

Low cost high power Yb or Nd doped crystals and fiber lasers with operating wavelength near 1 μm are commercially available and very popular for industrial applications [1–3]. Therefore, the research on near 1µm high power ultrafast laser has attracted a lot of attention due to its various industrial and medical applications, e.g. supercontinuum generation [4], nonlinear optics [5], ultrafast spectrographs, materials processing, graphene, thin film solar cells [6,7] and eye surgery. The generation of the passive ultrafast pulse is typically achieved by using the Semiconductor Saturable Absorber Mirror, SESAM, but its complicated fabrication process involves expensive Metal Organic Chemical Vapor Deposition MOCVD technique [8]. Thus there are strong research interests in developing some simple fabrication method for the carbon based saturable absorbers, such as carbon nanotube, CNT [9,10] and graphene oxide [11] that can effectively generate high power mode locking laser. Compared with the SESAM based absorbers, SWCNT has the advantages such as low cost of materials and fabrication process. Additionally, it has relatively broad operational waveband with demonstrated wavelength at 1μm [9,10], 1.3 μm [12], 1.5 μm [13], and 2 μm [14], respectively, as well as fast recovery time and great chemical stability. So far, reflective or transmission type SWCNT absorber has been fabricated by spin coating [10], spray coating, vertical evaporation coating methods [9,15]. However, it is difficult to ensure uniform deposition of CNT over a large area via spin / spray coating methods. Additionally these two methods required host materials, such as Poly(methyl methacrylate) (PMMA), sol-gel glass etc. to hold CNT together, and that will limit the output power of the CNT based mode locked seed laser system and some demonstrated records is below 50 mW [10,15] due to the lower melting point of the host material and the optical transparent of these absorber is highly sensitive to the solidification process, e.g. chemical gelatin, of the host materials, which will increase optical losses and degraded the laser performance.

Vertical evaporation method has disadvantages such as long fabrication time, raw material waste and so on. Dip coating proposed in this study is a main stream technique for industrial fabrication of large area thin-film materials. High quality thin film with large surface area, controlled thickness, good uniformity and continuity can be produced by this proposed low cost method [16]. Compared with the natural vertical evaporation deposition, more uniform thin film with more controlled quality can be produced via this method. The alignment direction of the coating material, coating thickness can be controlled by adjusting various coating parameters e.g. drawing direction, drawing speed, concentration etc. In this study the high quality transmission type SWCNT absorber was fabricated by dip coating method for the first time. The fabricated absorbers have a regular pattern and the SWCNT is aligned parallel with the withdrawal direction under the observation of Field Emission Scanning Electron Microscopy (FESEM) and Atomic force microscopy (AFM). High average mode locking laser power up to 2.7 W was achieved with 1064 nm operating wavelength, 167MHz repetition rate and 3.1 ps pulse duration by using this fabricated absorber.

2. SWCNT saturable absorber fabrication

In this experiment, the SWCNT used is purchased from Sigma Aldrich Inc. The amide functionalized SWCNT can be readily dissolved in DI-water, with about 5nm mean diameter of the SWCNT. The SWCNT powder was added into DI water to form a SWCNT-DI suspension with concentration of 0.03wt%, which is the maximum to avoid coagulation of SWCNT grains. Then the SWCNT-DI suspension was sonicated over 6 hours in a 400W ultrasonic cleaner, followed by adding 0.1 wt % Sodium dodecyl sulfate (SDS) and sonicating for another 1 hour. Then the suspension was centrifuged at the speed of 9500rpm for 20 mins to remove SWCNTs with large bundles before collected for the dip coating. The quartz plate substrate used is cleaned with ethanol, acetone and then sonicated. Finally substrate is treated with O₂ plasma to make the surface ultra-hydrophilic before it is ready to be used for dip coating. As experimental setup shown in Fig. 1(a), the quartz plate was immersed into the 0.03wt% SWCNT-DI suspension for withdrawn. The withdrawn process is controlled by a programmable dip coater. A thin suspension liquid film was coated on the substrate as shown in Fig. 1(b). As this liquid thin film was drew up and evaporated, it generated a flux to drive the coating material, SWCNT to the substrate surface and allows self-assembling to form the solid pattern [16]. The thickness of the liquid thin film, which is controlled by the drawing speed, liquid surface tension, temperature, relative humidity, etc., determined the thickness and regularity of the SWCNT solid film produced.

 

Fig. 1 (a) The schematic diagram of the mechanism of dip coating process. (b) Schematic diagram shows the self-assembly process in either vertical evaporation or dip coating method.

Download Full Size | PDF

For the non-colloidal solution, the suitable drawing speed tends to be higher in the range from 100µm/s to 1mm/s as self-assembly of the materials is not involved. However, the SWCNT-DI suspension is colloidal solution, slower withdrawn speed is therefore needed for self-assembly of the SWCNT. The water molecules within the SWCNT-DI solution would be pulled up and evaporated to drive and trapped the SWCNTs into the liquid thin film zone, in which the SWCNTs is self-assembled and then dry up to form a regular solid film pattern. The strips pattern observed from the absorber is due to the gravitationally driven jump mechanism repeats periodically [17]. Withdrawal of the substrate and evaporation from the container raises the height of the meniscus relative to the substrate surface. Eventually, the meniscus becomes too heavy. It tears off and slides to a new position determined by the receding contact angle of suspension on the substrate. The time required for a meniscus to form, stretch, and jump determines the stripe spacing [17]. Thus it is also very important to synchronize evaporation rate with the speed of the formation of the liquid film, which is controlled by the drawing speed. Fringe separation would be too large and no continuous film was formed for drawing speed at 30nm/s and 63 nm/s as Figs. 2(a) and 2(b) shown. Since the estimated laser spot incident on the absorber is around 120µm in diameter, the fringe with large separation in the range of tens of micron meter therefore would not be able to produce stable CW mode locking pulses as results shown in Table 1.

 

Fig. 2 The optical microscope image of the formed SWCNT film with dipping condition of (a) 30nm/s, (b) 63nm/s and (c) 96nm/s withdrawn speed, 22 °C dipping environment temperature, 40% relative humidity and 0.03wt% SWCNT-DI suspension concentration with 0.1wt%SDS being added. (Due to the reflection of the light, the dark region has less density of SWCNT.)

Download Full Size | PDF

Table 1. The dip coating samples produced with various parameters and lasing performance. Con. film –continuous film; UQML-unstable Q-switch mode lock; SCWML-stable CW mode locking. The temperature and relative humidity were kept to be constant at 22 °C and 40%, respectively.

View Table

As shown in Fig. 2(c) and Table 1, the uniformity and continuality of the SWCNT film and its laser performance has been significantly improved by increasing drawing speed up to 96 nm/s (sample c). Sample C absorber has been used within the laser cavity to produce stable CW mode locking as shown in Table 1. However, with the increase of the drawing speed up to 160 nm/s, the produced SWCNT absorber has no significant changes in film pattern and optical transmission, which indicates that the optimal drawing speed is close to ~100 nm/s for these conditions. We expected the uniformity of the SWCNT film can be further improved by optimizing drawing temperature, humidity and more stable environment.

The sample D was produced by reducing suspension concentration to 0.015wt%, but all other dipping parameters are remained unchanged compared with that of sample C. Then the initial transmission ratios of fabricated SWCNTs-SA samples are measured by using FTIR as shown in Fig. 3(c) and Table 1. The absorption of the Sample D at 1064 nm reduces as expected when compared with sample C. Under same drawing conditions and evaporation rate of the dip coating process, the lower concentration of the SWCNT-DI suspension leads to less CNT being trapped in the liquid thin film, thus eventually fabricating a SWCNT solid film with lower absorption. In this way, the film thickness and initial transmission of the SWCNT absorber can be changed but its film pattern and continuity remains unchanged. Figure 3(a) and 3(b) show the alignment of the SWCNTs on Sample D are parallel with the withdrawn direction. It is proved that the SWCNT alignment direction can be controlled via the dip coating method due to the self-assembly effect, further enhancing the regularity and uniformity of the absorber, which is a unique feature when compared with absorber fabricated by other methods e.g. spin or spray coating methods. This well aligned SWCNT can serve as optical polarizer [18]. That may be potentially used to produce polarized laser output. It takes only 15 hours to produce a 2cm x 2cm continuous SWCNT thin film on a quartz substrate. In principle, the production yield can be easily enlarged by using wider substrate. The optimal drawing speed is determined by evaporation rate, surface tension of substrate and liquid, speed of self-assembling of SWCNT. Therefore, the sample drawing speed can be further increased by adjusting temperature and humidity. The remaining suspension can be dilated and recycled for the next dip coating process, indicating its great potential for mass production.

 

Fig. 3 (a)The FESEM, (b)AFM image shows the surface topography of the Sample D; (c) The Optical transmission spectra of the SWCNT-SA samples fabricated by various dipping coating conditions as shown in Table 1.

Download Full Size | PDF

3. Laser experiments and discussion results

The SWCNT absorber inserted about 3mm apart from the end mirror to high incident intensity on the absorber as illustrated in Fig. 4(a). Figure 4(b) shows the average output power with respect to various incident pump power and respected operational modes. The threshold for stable CW mode-locking and slope efficiency of Sample C are 7.93W and 34%. The threshold for stable CW mode-locking and slope efficiency of Sample D are 4.5W and 39%. Due to the lower insertion loss of Sample D, the overall mode locking performance by using Sample D is better. Its maximum average output power achieved is up to 2.7W. Figure 5(a) shows the pulse train recorded for the continuous wave mode locking, CWML, operation of Sample D under 8.36 W pump power, with the repetition rate of the pulse train close to the round-trip cavity time, indicating successful CW mode locking achieved. The autocorrelation trace shown in Fig. 5(b) is obtained by using an Intensity Autocorrelator (Femtochrome Research Inc. FR 103-XL). The measured FWHM of pulse width is 3.1 ps for the 2.7 W maximum average output power from the mode locking system of using Sample D. The repetition rate is 167 MHz. The estimated pulse energy and peak power are 16.1 nJ and 5.2 kW, respectively.

 

Fig. 4 (a) The schematic diagram of laser, HR-High reflection, T-Transmission ratio, RoC – Radius of Curvature of the mirror. (b) The Average output power at 1064nm verse 808nm pump power for CW and mode-locking operation.(iQSML - irregular Q-switch mode locking mode; RQSML - regular Q-switch mode locking mode; CWML - CW mode locking mode).

Download Full Size | PDF

 

Fig. 5 (a) The pulse train of the CWML operation with output power of (b)The corresponding Autocorrelation spectra with respect to pump power of 8.36W. (c) The nonlinear transmission of the SWCNT-SA sample D with excitation wavelength at 1064 nm. Dots: measured data; solid curve: fitting to the data.

Download Full Size | PDF

In our previous publication [9], we produced mode-locked laser with 15.7 ps pulse width using SWCNT saturable absorber fabricated by vertical evaporation method, which is longer than our current system with produced pulse width of 3.1 ps. The different pulse widths produced in these two experiments can be attributed to various reasons, including the different average diameter of the SWCNT, alignment patterns, absorption curve, modulation depth and saturation fluence. In fact, shorter pulse width produced is expected by using saturable absorber with larger modulation depth [19]. In this experiment the measured modulation depth is 8% as shown in Fig. 5(c), which is higher than the modulation depth of the previously used absorber, which is only 4.4% [9].

4. Conclusions

SWCNT saturable absorbers were fabricated by using dip coating method under various conditions. The saturable absorbers features e.g. initial transmission ratio, film pattern was optimized by adjusting different withdrawn parameters, e.g. drawing speed, SWCNT concentration. The demonstrated fabrication method is simple, scalable, cost effective and much more suitable for industrialization. These absorbers have been successfully used within a diode pumped Nd:YVO₄ mode locking laser system operating at 1064 nm to produce high output power up to 2.7 W with continuous wave mode locking pulse duration and repetition rate of 3.1 ps and 167MHz. The calculated pulse energy and peak power are 16.1nJ and 5.2kW, respectively.