1. Introduction

Holographic volume Bragg gratings (VBGs) have a great potential as useful devices for fiber optic communication systems such as tunable filters, notch filters, wavelength-division demultiplexers, and grating couplers, because of its use to make a narrow bandwidth of holographic filters. Numbers of studies introduced VBG structured holographic filters using various types of holographic recording materials such as photorefractive crystals, dichromated gelatin, photopolymers, and photosensitive glasses [1]. Among those materials, photopolymer is the one of attractive materials for making VBG that provide high diffraction efficiency, dry fabrication processing, and low price. However, because of the limitation in thickness of most available photopolymers, the feasible wavelength bandwidth of fabricated VBG is still broad. For example, the bandwidth of conventional VBG using commercially available DuPont photopolymer films with few tens of micrometers thick is at several tens of nanometers. The thickness limitation influence filter performance makes it difficult to meet the spectral bandwidth requirements for wavelength-division multiplexing (WDM) optical communication systems. In this paper, we introduce a design approach and fabrication procedures of a new type holographic filter with a CVBGs structure embedded in photopolymer waveguide film. For recording CVBGs, a 20μm-thick DuPont photopolymer HRF600x123-20 is adopted because they used for recording transmission and reflective hologram with high refractive index modulation over a broad range of grating periods [2].

2. Structure and operation characteristics

The schematic diagram of the proposed holographic filter is illustrated in Fig. 1 . The filter is composed of two VBGs; a slanted VBG with a grating inclination angle $\phi ={45}^{\circ }$ (${45}^{\circ }$-VBG), a non-slanted VBG with a grating inclination angle $\phi =0^{\circ }$ ($0^{\circ }$-VBG), where the grating inclination angle is the angle between the grating vector ${\text{K}}_1$ (or ${\text{K}}_2$) and the z axis as shown in Fig. 1. Those VBGs are embedded in a photopolymer film with thickness $t_g$ wherein, two gratings are separated by a waveguide of length $z_L$ along the direction of beam propagation. $I_{inc}$, $I_t$, and $I_{out}$ denote the intensities of the incident input beam, the transmitted beam, and the output reflected beam, respectively. $z_o$ is the position of the incident input beam on the surface of the ${45}^{\circ }$-VBG, $w_g$ is the width of the gratings, and the $d_{g1}$ and $d_{g2}$ are the grating lengths for the ${45}^{\circ }$-VBG and the $0^{\circ }$-VBG, respectively. In this scheme, if the refractive index of the photopolymer film is higher than that of the surrounding medium, the photopolymer film acts as both of the grating medium and the waveguide layer to guide diffracted beam between the gratings. We name this type of filter “a waveguide-CVBGs filter” according to its structure. In the configuration of Fig. 1, the ${45}^{\circ }$-VBG is operating both as an input and output coupler [3–5]. By this ${45}^{\circ }$-VBG, a nearly surface-normal-incident input beam is ${90}^{\circ }$ diffracted into the plane of the waveguide and guided toward the $0^{\circ }$-VBG. The guided beam is then retro-diffracted by $0^{\circ }$-VBG in the opposite direction, and again out-coupled by the ${45}^{\circ }$-VBG. The second $0^{\circ }$-VBG is working as a reflective grating which diffracts the guided beam back into the input/output coupler (${45}^{\circ }$-VBG). Considering the functions of two gratings and the structure of device, we see that the spectral bandwidth of a waveguide-CVBGs filter is controlled not by adjusting film thickness but by the effective interaction length $d_{\mathit{eff}}$ between the $0^{\circ }$-VBG with a grating length $d_{g2}$ and the guiding beam in photopolymer waveguide. Therefore the effective grating length from the guiding beam standpoint functionally substitutes for the thickness of the $0^{\circ }$-VBG. In this structure, the two gratings are cascaded with different regions inside a single waveguide, the performance analysis of the waveguide-CVBGs filter becomes complicated. But still the angular and spectral characteristic of the filter may be dominated by the Bragg effect, since the filter is composed of a thick volume Bragg gratings with a wide and thick waveguide. In our proposed structure, the two VBGs are formed as transmission geometry and reconstructed in reflection mode of operation by surface-normal illumination on the ${45}^{\circ }$-VBG. In Fig. 1, the grating vectors ${\text{K}}_1$ (for ${45}^{\circ }$-VBG) and ${\text{K}}_2$ (for $0^{\circ }$-VBG) have magnitudes of $\left|{\text{K}}_1\right|=2\pi /{\mathrm{\Lambda }}_1$and $\left|{\text{K}}_2\right|=2\pi /{\mathrm{\Lambda }}_2$ respectively, where ${\mathrm{\Lambda }}_1$ and ${\mathrm{\Lambda }}_2$are the grating periods. The grating periods of the two VBGs, each with a different periods, can be expressed separately from the Bragg condition [6] as ${\mathrm{\Lambda }}_1={\lambda }_c/2n_{gc}\cos ({45}^{\circ }-{\theta }_{c1})$ and ${\mathrm{\Lambda }}_2={\lambda }_c/2n_{gc}\cos {\theta }_{c2}$, where ${\lambda }_c$ is the Bragg-matched coupling wavelength of the incident input beam, $n_{gc}$ is the average refractive index of the grating, ${\theta }_{c1}$ and ${\theta }_{c2}$ are the coupling angles of the incident beams for each ${45}^{\circ }$-VBG and $0^{\circ }$-VBG in the gratings, respectively. In case of $n_{gc}=1.49$, ${\theta }_{c1}=0^{\circ }$, ${\theta }_{c2}=0^{\circ }$, we have a different grating periods ${\mathrm{\Lambda }}_1=~0.735\mu m$ and ${\mathrm{\Lambda }}_2=~0.520\mu m$ at 1550nm coupling wavelength.

 

Fig. 1 Schematic diagram of a waveguide-CVBGs filter.

Download Full Size | PDF

3. Fabrication details

In this paper, two gratings are recorded by using a same Nd:YAG laser source with free-space wavelength ${\lambda }_w=532nm$. To write the different gratings with different optical setups, the laser beam is divided into two beams with a beamsplitter, and one beam is selected for constructing of the $0^{\circ }$-VBG recording setup. The other beam used for constructing of the ${45}^{\circ }$-VBG recording setup. Then the two laser beams are spatially filtered and then collimated by same optical components to obtain an identical collimating beam with a uniform phase–front for each setup. The each dimension of the collimated recording beams are$7mm\times 7mm$, and TE-polarized such that the electric field is perpendicular to the plane of incidence. The photopolymer is laminated onto a fused-silica glass substrate with dimensions of $85mm\times 25mm\times 1.5mm$ after one of the Mylar cover sheets is removed. Next, we recorded the $0^{\circ }$-VBG inside the photopolymer, and then moved the sample to the ${45}^{\circ }$-VBG recording setup position as shown in Fig. 2 . The $0^{\circ }$-VBG is recorded by the conventional symmetrical recording geometry. In the experiment, the half angle of the recording beam in the air was${30.987}^{\circ }$, and each recording beams had an equal intensity of $0.12mW/c{m}^2$ and exposure energy density was $49mJ/c{m}^2$. However, in the asymmetrical recording geometry as the ${45}^{\circ }$-VBG, the material shrinkage is predominately in the normal direction to the surface of the grating, and it changes the grating slant angle and period. The measured shrinkage of the DuPont photopolymer HRF-600X123-20 with $t_g=20\mu m$ for a ${45}^{\circ }$-VBG was ~4% after UV fixing processing. In this case, the recording angles of the ${45}^{\circ }$-VBG in Fig. 2 with considering material shrinkage in air are calculated to be ${\theta }_{wc1}={27.215}^{\circ }$ and ${\theta }_{wc2}={16.426}^{\circ }$ respectively [5]. Unlike conventional recording setup for a $0^{\circ }$-VBG, the interferometric recording architecture including a prism system as shown in Fig. 2 is applied in order to fabricate the input/output couplers with a ${45}^{\circ }$ slanted volume phase grating. Before recording the ${45}^{\circ }$-VBG, the sample was sandwiched by the two identical fused-silica ${45}^{\circ }-{45}^{\circ }-{90}^{\circ }$ prisms, and the refractive index matching oil which is inserted between the input prism and substrate, between the cover sheet and the output prism. After that, the sandwiched sample is placed on a rotation-translation stage, the rotation-translation stage is mounted on a 2-axis translation stage to arrange accurately the recording beams at the desired position inside photopolymer film. The dimensions of a fused-silica prism were of $60mm\times 60mm\times 25mm$, and the prisms were antireflection coated to reduce reflections. The refractive-indices of the photopolymer film are $n_{gw}=~1.51$ and $n_{gc}=~1.49$ at ${\lambda }_w=532nm$ and ${\lambda }_c=1550nm$. In addition, the glass substrate material is a fused-silica like that of prisms with refractive index $n_{pw}=1.460$ and $n_{pc}=1.444$ at ${\lambda }_w=532nm$ and ${\lambda }_c=1550nm$. To attain a high contrast ratio of the recording interference pattern through the compensation of the incident powers for the different recording angles, the intensity of the two recording beams are adjusted to be $I_{wc1}=0.15mW/c{m}^2$ and $I_{wc2}=0.081mW/c{m}^2$ respectively. The dimensions and the separation length of the fabricated CVBGs are $~8.6mm(d_{g1})\times 7mm(w_g)$ for a ${45}^{\circ }$-VBG, $~8mm(d_{g2})\times 7mm(w_g)$for a $0^{\circ }$-VBG, and 3mm ($z_L$), respectively.

 

Fig. 2 Experimental setup for fabricating ${45}^{\circ }$-VBG.

Download Full Size | PDF

4. Experimental results and discussion

After recording the two gratings, the sample is fixed by UV light. The fixed sample is then mounted onto a rotation-translation-goniometer stage for performance testing as shown in Fig. 3 . The spectral dependence of our device is measured using the Newport’s external cavity tunable laser with a center wavelength of 1550nm with tunable wavelength range of ± 30nm. The dimension of the incident laser beam at the surface of the sample is approximately$2mm(z)\times 5mm(y)$. The polarization of the incident laser beam is controlled by a $\lambda /2$ plate and a linear polarizer. For the experiment the TE polarized incident beam was adopted, since waveguide-CVBGs filter is polarization-sensitive device with strong reflection efficiency for it [6]. The wavelength of the tunable laser is monitored using an optical spectrum analyzer and the intensity of the transmitted and the reflected output beams are measured through a detector. The intensity profile of reflected output beam is obtained by replacement of the detector with an infrared (IR) camera. It is noted that the proposed filter is very angularly-spectrally sensitive device to have combined properties of the planar waveguide and the volume Bragg grating. In Fig. 3, angular tuning is accomplished by rotating the sample about y-axis perpendicular to the sample’s facet. In this procedure, angular tuning at a given wavelength may lead to a wave-vector mismatch among the gratings and the incident and diffracted beams inside the sample. As the light of wavelength to satisfy the Bragg conditions of two gratings is diffracted, the precise angular and wavelength tuning are necessary. The adjusting procedure for an optimized coupling starts with the first step for tuning a tunable laser to the peak reflectance wavelength for the nearly normal incident input beam. The incident angle is then varied to the coupling angle having a peak reflectance, and then fine-tunes again the wavelength and the incident angle around the obtained values from the first step to confirm the optimizing coupling.

 

Fig. 3 Experimental setup for testing the fabricated waveguide-CVBGs filter.

Download Full Size | PDF

In this paper, we define the output reflectance of the filter as $R=(I_{out}/I_{inc})\times 100\%$ and the transmittance of the ${45}^{\circ }$-VBG as $T=(I_t/I_{inc})\times 100\%$ in Fig. 3. Where the incident intensity of input beam $I_{inc}$ denotes the intensity of the transmitted beam through the film region without grating at input coupling angle with considering the Fresnel and absorption losses, $I_{out}$ is the intensity of the out-coupled output beam from the a ${45}^{\circ }$-VBG in air, and $I_t$is the intensity of the transmitted beam through the ${45}^{\circ }$-VBG.

Figure 4 shows the angular response of the normalized reflected intensity for a fixed coupling wavelength of 1547.85nm obtained by the above procedure. The input beam is positioned approximately at $z_o=5mm$ on the surface of the sample as shown in Fig. 1. As shown in Fig. 4, the measured input coupling angle was ${\theta }_{c1}={1.301}^{\circ }$ in the grating. It is also noted that the response curve decreased slightly asymmetrically at the central wavelength unlike the conventional uniform VBGs, with a full bandwidth of approximately ${0.108}^{\circ }$ at the 98% decreased level of the peak response. The measured angular selectivity at the full width at half maximum (FWHM) inside the grating was about ${0.026}^{\circ }$. Figure 5 shows the transmittance spectrum of the ${45}^{\circ }$-VBG for $z_o=2mm$, $5mm$ at the fixed input coupling angle of ${\theta }_{c1}={1.301}^{\circ }$. As seen in Fig. 5, each dip in the transmittance spectrum corresponds to the Bragg diffraction of incident light into a waveguide mode supported by the ${45}^{\circ }$-VBG, where each mode is associated with a specific coupling angle. It is noted that the measured Bragg coupling wavelength of the grating at the input coupling angle corresponds to the wavelength at which a maximum dip in the transmittance spectrum. The measured transmitted efficiencies at the coupling wavelength of 1547.85nm are 38% (i.e. the input coupling efficiency of ${45}^{\circ }$-VBG is 62%) for the $z_o=2mm$ and 26.7% for the $z_o=5mm$. To our knowledge, suitable theories are not available to model the spectral response of our waveguide-CVBGs configuration. Therefore, experimental results are analyzed and compared approximately by use of the Kogelnik’s coupled-wave theory [7] without considering the effects of waveguide although it’s not accurate representation of the spectral responses of transmitted and reflected beams. Figure 5 also includes a simulation curve with the coupled-wave theory for a reflection grating at $t_g=20\mu m$, $\phi ={45}^{\circ }$, ${\lambda }_c=1547.85nm$, and refractive index modulation $\mathrm{\Delta }n$ = 0.0047. The refractive index modulation was obtained by using the measured reflection efficiency of the ${45}^{\circ }$-VBG for the $z_o=5mm$. If we consider only the curve around the wavelength of 1547.85nm by excluding the other guided modes in Fig. 5 as a conventional reflection grating, a measured wavelength bandwidths at the FWHM level for each curve at the coupling wavelength are ~1.7nm for the $z_o=2mm$ and ~1.3nm for the $z_o=5mm$. It can be seen that the experimentally measured bandwidth is smaller than the theoretical value (~2nm), also the shape of experimental curves are more complicated because of the waveguide effect and the additional mode excitation resulting from the discontinuity between the waveguide and the VBGs.

 

Fig. 4 Normalized reflected intensity versus the (internal) incident angle for a fixed coupling wavelength.

Download Full Size | PDF

 

Fig. 5 Spectral response of the transmittance of the ${45}^{\circ }$-VBG for the fixed input coupling angle.

Download Full Size | PDF

Figure 6 shows the spectral response of the output reflected efficiency as a function of tuning wavelength in the range between 1538.60nm to 1549.35nm for the fixed input coupling angle of ${\theta }_{c1}={1.301}^{\circ }$ and the positions of the incident input beam of $z_o=2mm$, $5mm$. A measured wavelength bandwidth at the FWHM level is ~0.45nm for each curve of centered at 1547.4nm, and output reflectance R is ~9.7% for a $z_o=2mm$ and ~14.17% for a $z_o=5mm$ respectively. The Bragg coupling wavelength for the maximum dips in Fig. 5 are in agreement with those depicted in Fig. 6, indicating that other waveguide modes except the maximum dips does not nearly affect the reflectance spectrum because of the Bragg matching condition. Therefore, we can see that the spectral response in the case of our configuration is mainly due to the $0^{\circ }$-VBG, the spectral bandwidth of the $0^{\circ }$-VBG for the ${\theta }_{c2}\approx 0^{\circ }$ can be determined approximately from the coupled-wave theory as $\delta {\lambda }_{0^{\circ }-VBG}\cong {\lambda }_c{}^2/2n_{gc}{d}_{\mathit{eff}}$. Using the measured values with ${\lambda }_c=1547.85nm$ and $\delta {\lambda }_{0^{\circ }-VBG}\cong 0.45nm$, we can calculated the effective interaction length $d_{\mathit{eff}}$ of $~1.8mm$. The calculated $d_{\mathit{eff}}$ is shorter than the grating length of the $0^{\circ }$-VBG, which is due to the fact that more of light is diffracted from the area closer to the input boundary of the $0^{\circ }$-VBG and corresponds to the features of a uniform volume grating [3]. Additionally, Fig. 6 shows a fit of the coupled-wave theory to the measured data for the $z_o=5mm$ with a fit parameters$\mathrm{\Delta }n=1.09\times {10}^{-4}$, ${\lambda }_c=1547.85nm$, and $d_{\mathit{eff}}=1.8mm$. The output reflectance with respect to the positions of incident input beam $z_o$ on the surface of the ${45}^{\circ }$-VBG were investigated. We observed from the measurement results, as $z_o$ gets closer toward the grating-waveguide boundary, the output reflectance is increased. This is because of that as the incident-beam position approaches to the grating-waveguide boundary, the leakage of the guided mode from the waveguide decreases, which is good agreement with the results of output-coupling effect in Ref. 4. On the other hand, strongly coupled few modes are appeared as seen in Fig. 6 with the incident beam approach to near $z_o=d_{g1}$, because of the waveguide effect similar to the input/output coupler formed inside the photopolymer waveguide [8]. Therefore, it is seen that the position of incident input beam $z_o$ should be limited to a certain position within $d_{g1}$ to avoid the unwanted waveguide effect and to optimize the performance in the device.

 

Fig. 6 Normalized reflected intensity versus wavelength for the fixed input coupling angle.

Download Full Size | PDF

Figure 7 shows the intensity profile of the output beam for the incident conditions ${\theta }_{c1}={1.301}^{\circ }$, ${\lambda }_c=1547.85nm$and $z_o=5mm$, where the reflected output beam is captured by an infrared (IR) camera with a mounted zoom lens in Fig. 3. In Fig. 7, the angular separation of the two beams caused by the angle difference between the output beam and the back-reflected beam (left spot in Fig. 7) of the incident beam at the surface of the sample is observed. The shape of the output diffracted beam is a typical aspect for the ${45}^{\circ }$-VBG with relatively long grating length as our rectangular-shaped grating configuration. The diffracted beam spreading can cause a significant loss as coupling the output beam inside a fiber however, it can be minimized by the reducing the grating length of the ${45}^{\circ }$-VBG and also by altering the shape of the incident beam such as a collimated Gaussian beam.

 

Fig. 7 Photograph of the intensity profile of the reflected output beam obtained by the IR camera.

Download Full Size | PDF

In our experimental results, the measured input coupling angle and wavelength are shifted from the design values ${\theta }_c=0^{\circ }$ and ${\lambda }_c=1550nm$, that are primarily due to the arrangement errors in optical setups. Moreover the output performance of the filter is sensitive to the throughout alignment from the fabrication to the measurement procedures. This can be corrected by more accurate alignment and recording, and testing processes. In addition, one of the most significant drawbacks associated with the DuPont photopolymer film have a nonplanar surface is its high attenuation when used it as a waveguide. The reported waveguide loss, measured by the prism coupling method at a wavelength of 850nm, was approximately 5.5dB/cm [5], therefore holographic recording media with a low propagation loss and planar waveguide structure are critically needed to practical applications. Recent developments in holographic materials have made it possible to fabricate our special configuration in a planar waveguide type. One method to overcome this problem is using low-loss photopolymer material coated on the glass or photosensitive glass materials [9], VBGs can be fabricated directly as waveguide type in these media, leading to possibly better performance than the used type in this paper. Moreover, two VBGs in our experiments have constant amplitude over the grating regions, therefore the spectral profile in Fig. 6 is typical feature of the photopolymer(or photorefractive crystal)-based filters. However, the flat-top profile is highly desired for WDM system applications, which is expected through a hologram apodization method [10].

5. Conclusions

In this paper, a novel holographic filter operating at 1550nm wavelength region using a waveguide-CVBGs structure was proposed and experimentally demonstrated by using a 20 μm-thick DuPont photopolymer film HRF600x123-20. Detailed fabrication and testing procedures and a performance characteristic were also presented. The fabricated holographic filter had a peak reflectance of ~14.17% with a FWHM of ~0.45nm at center wavelength of 1547.85nm. It is shown that the CVBGs configuration embedded in a photopolymer waveguide film is potentially useful mean to overcome the limitation by the thickness of photopolymer used, to attain narrow spectral bandwidth characteristic without increasing the film thickness. We believe that our proposed the waveguide-CVBGs structure is widely applicable to integrated optics and WDM optical communications for narrower spectral bandwidth characteristics.