1. Introduction

Following the emergence and the rapid progress in the field of nanophotonics, there is a growing need for accurate characterization techniques with nanoscale resolution. Indeed, techniques such as stimulated emission depletion (STED) microscopy [1], and single molecule microscopy [2,3] have been evolved to become high impact imaging characterization approaches, as evident by awarding the recent Nobel Prize to the inventors of these methods. In parallel, near field scanning optical microscopy (NSOM) [4, 5] is rapidly developing, and it is now being used in a wide range of applications [6–16]. While there are several different NSOM configurations in use, one would typically differentiate between aperture and apertureless NSOM. In the first approach, the near field is typically collected by a dielectric probe (e.g. pulled optical fiber) which may be coated by metal. The collected signal is guided to a photodetector. This approach requires a fairly simple experimental setup, but its spatial resolution is limited because of the aperture size which is ultimately limited by the skin depth of the metal and by signal to noise ratio considerations. On the other hand, the apertureless NSOM may provide enhanced spatial resolution at the expense of a more complex experimental setup and background noise.

In recent years, following the miniaturization of nanophotonic devices, light induced heating (i.e. self-heating of nanophotonic devices as a result of light absorption within the device) is becoming a crucial issue. This is particularly true for high speed active silicon photonic and plasmonic devices, in which effects such as free carrier absorption and ohmic losses (Joule heating) play a crucial role. While typically considered as an undesired effect, light induced heating can also play a positive role, and it was already shown that light induced heating of plasmonic structures e.g. nanoparticles can be used for the heating of liquids [17,18], magnetic data storage [19,20], micro-fluidics [21,22], medicine [23,24], and nano-chemistry [25]. Considering the importance of thermal effects in nanophotonic devices, there is a growing need for characterizing temperature distributions within such devices at the nanoscale. The Scanning Thermal Microscopy (SThM) is a viable candidate for mitigating this challenge, as it can provide nanoscale spatial resolution with thermal resolution in the milli Kelvin regime [26–29]. Indeed, it was already shown [30,31] that SThM can be used for thermal profile characterization of light induced heating in silicon photonics and plasmonic devices at the nanoscale.

Recently, it was shown that on top of measuring the temperature profile of nano photonic devices, SThM can also be used for the purpose of measuring the near field of light interacting with nanophotonic devices [32,33]. Following these demonstrations it is now even more important to understand the capabilities of the SThM approach in measuring light and temperature distributions within nanophotonic devices. This is the main goal of the current manuscript. To address this goal, we perform SThM characterization of photonic devices made of several absorbing and non absorbing material platforms (doped silicon, undoped silicon and silicon nitride), and study the frequency response of our system. Additionally, we provide SThM measurements at several wavelengths spanning over an octave. Furthermore, we perform a case study to compare the mechanisms of light measurements versus temperature measurements.

Our results show that in the context of dielectric nanophotonic devices, the SThM is responsive mostly to the optical near field profile via the mechanism of light absorption in the SThM probe, rather than to light induced heating effect in the nanophotonic sample. As such, the SThM approach may provide a simple, yet robust, technique for optical near field characterization of large variety of nanophotonic devices over a very broad spectral range without the need for advanced optical setups. On the other hand, it may fail in achieving its original purpose of measuring the temperature profile within the nanodevices. To address this latest issue, we also propose and study a configuration which may allow the SThM to serve its original purpose of measuring the temperature profile of nano photonic devices by using a spacer layer and taking advantage of the differences in spatial distribution of the thermal and the optical mode.

2. SThM measurement of optical near field

To study the SThM capabilities in measuring the optical near field properties of nanophotonic waveguides and resonators we use the following setup, as shown in Fig. 1. Briefly, light is coupled into and out of a nanoscale silicon waveguide using a lensed fiber. The incident light is derived from a tunable laser diode source at the telecom band around the wavelength of 1550nm (Agilent 8164a). Typically, the optical power within the waveguide is ~10µW. The output fiber is connected to a photo detector for the purpose of monitoring the light coupling efficiency as well as for tracing the resonant frequencies of our resonators. In parallel, the device is scanned using an SThM probe (operating in tapping mode), which is a thermocouple made of a Pt-Au junction (apex diameter of ~300nm, Nanonics Ltd). To increase the SNR of our system, the output of the SThM is connected to a lock-in amplifier (LIA, Stanford SRS 830). The LIA is locked on the laser modulation frequency (700 Hz, obtained using internal laser modulation) and the output of the LIA is connected to a computer. We explicitly note that there is no optical detector in the SThM measurement setup.

 

Fig. 1 Schematics of the experimental setup. Lensed fibers are used to couple light into and out of a silicon nanowaveguide. In parallel, the SThM probe is scanning the waveguide and the resonator. The signal is recorded by a computer. The insert (right of the Micro-Ring resonator) provides SEM images of the SThM probe.

Download Full Size | PDF

The first nanophotonic device that was measured using the above mentioned setup is a Boron doped$(5x{10}^{17}c{m}^{-3})$ silicon nano waveguide (220nm height and 450nm wide). Two major heat generation mechanisms are expected to play a role in such a device: free carrier absorption (FCA), which is mainly the result of the high doping level, and two photon absorption (TPA) which is expected to be less dominant given the relatively low power levels within the structure. The results of the SThM measurements can be seen in Fig. 2(a) and Fig. 2(c). One can clearly observe two main beating periods - one with a period of ~0.31µm and another one with a period of ~2.1µm. The short period (0.31µm) is originated from the interference of the counter and co-propogating fundamanetal modes. The longer period (2.1µm) is originated from the interference of the fundamental and the higher co-propagating modes. These two measured beating periods agree very well with the calculte effective indices (Comsol multiphysics Ltd) of the two modes, $n_{eff}$ = 2.52 and 1.83, for the fundamental and the higher order mode respectively.

 

Fig. 2 (a,b) - Thermal (SThM) and NSOM signals collected from a doped silicon nano waveguide, respectively. The waveguide is bended, with a bending radius of 50 µm. (c,d) Magnified image of panels a,b respectively.All scale bars are in arbitrary units.

Download Full Size | PDF

Considering the high heat conductivity of silicon, one cannot expect the light induced temperature profile of the waveguide to contain such small periods, because of the significant heat diffusion within the silicon (this was verified by a finite element heat 3D simulation). Therefore, we attribute the obtained signal to the direct measurement of the optical near field intensity profile of the nano waveguide, rather than to its light induced temperature profile. The optical near filed signal is originated from plasmonic enhanced light absorption in the SThM probe (see section 5). To further support our claim, we have repeated this measurement, this time with an aperture NSOM probe (opening diameter of ~250nm, Nanonics Ltd.) instead of the SThM probe. The results, shown in Fig. 2(b) and Fig. 2(d) are very similar to the SThM measurement, in support of our hypothesis.

While the comparison between SThM and NSOM measurement provides a clear indication for the responsivity of our SThM probe to the optical near field, one may still suspect that the absorption of light in the silicon nano waveguide is a major source for the obtained signal. In order to address this suspicion, we have repeated our measurements using a different set of samples in which the light absorbing mechanisms are diminished. Our samples of choice were: 1- undoped silicon and 2 - $S{i}_3{N}_4$waveguides and resonators. By using undoped silicon samples one can significantly reduce FCA, and the - $S{i}_3{N}_4$samples allow eliminating TPA in wavelengths around the telecom band. The obtained measurement results are presented in Fig. 3.

 

Fig. 3 SThM images of (a) undoped silicon waveguide, (b,c) undoped silicon micro-ring resonator off and on resonance respectively, (d) $S{i}_3{N}_4$micro-disk resonator at the wavelength of 980nm, (e,f) a section of a $S{i}_3{N}_4$ micro-disc resonator at wavelengths off and on resonance respectively. All images exept panel d were measured at the telecom band around 1550 nm. The green arrow represents the light propagation direction. The Q-factor of both resonators are ~30,000 @1550nm. Beating is clearly observed only in panel a because of the better spatial resolution used for this result (note the difference in scale bars).

Download Full Size | PDF

The results presented in Fig. 3 show a clear indication for our capability to measure the optical near field using the SThM approach. The absorption of light within these samples is negligible, yet a strong signal is still evident, which is attributed to the optical near field that is absorbed by the apex of the SThM probe.

3. Optical near field Vs. light induce heating measurements using SThM

These results, as well as previously reported results [32,33] provide a clear indication for the capability to measure the optical near field using the SThM approach. On the other hand, we have previously observed thermal signature of the light induced heating in both nano waveguides and plasmonic structures [30,31]. It is thus essential to differentiate between the two effects and to better understand under what circumstances one would expect the system to be mostly responsive to either light distribution or thermal distribution of a nanophotonic device. To do so, we experimentally measure the responsivity of the SThM probe in two extreme cases: 1) direct sample heating (e.g direct temperature measurement) with no presence of light, such that light absorption in the SThM probe is obviously eliminated and 2) direct light absorption inside the SThM probe without additional sample heating.

In the case of direct sample heating, an Au nanowire (80nm high, 400nm wide, 350µm long) on top of a silicon substrate was fabricated by lift off process. The nanowires were covered by a thin (~10 nm) layer of oxide in order to prevent electrical shortcut between the nanowire and the SThM probe. The nanowire was connected to a tunable voltage source. Due to the electrical power dissipation inside the nanowire $\left(Q\propto P=\frac{V_{in}{\left(t\right)}^2}{R}=\frac{{\left(V_{in}\cos \left(2\pi ft\right)\right)}^2}{R}=\frac{v_{in}^2}{2R}\left(1+\cos \cos \left(2\pi \left(2f\right)t\right)\right)\right)$, heat is generated and the increase in temperature is measured using the SThM probe [Fig. 4(a)].

 

Fig. 4 a) Schematics of direct heat measurement using Au nano wire. An AC voltage source is connected to the nanowire through two contact pads. b) Illustration of the SThM probe above the facet of an optical fiber for the purpose of directly measuring light distribution by the SThM probe.

Download Full Size | PDF

We start this measurement by performing an initial scan in order to locate the center of the nanowire. Then, we fix the position of the probe at this position as we change the input voltage (and consequently the incident electrical power) and record the output of the LIA as a function of the input power.

Next, we turn into measuring the direct light absorption inside the probe. To ease on this task, we have used an experimental setup as shown in Fig. 4(b). In this configuration, the SThM probe interacts with the optical beam propagating through the top facet of a cleaved optical fiber (SMF-28). First, an initial scan is performed in order to find the center of the optical mode in the fiber. Then, we fix the position of the probe at the center of the optical mode, change the incident optical power inside the fiber (at 1550nm wavelength) and record the output of the LIA as a function of the incident optical power.

After measuring the SThM signals for the cases of direct heating and direct light absorption inside the probe we can now compare the sensitivity of the SThM probe in both cases. This comparison is shown in Fig. 5, where the SThM signals as a function of the incident power (electrical or optical) are plotted.

 

Fig. 5 light induced heat measurement (blue) at the SThM probe for the wavelength of 1550nm and direct temperature heating (purple) measurements of the nanowire. The voltage source is modulated at 352Hz (and thus heat generation is modulated at double the frequency, i.e. 704Hz) and the laser is also modulated at 704Hz. The obtained slopes are $0.0298\pm 8\text{x}{10}^{-4}$ [nA/mW] and $0.0298\pm 8\text{x}{10}^{-4}$ [nA/mW] for the light induced heating and the direct heat measurements, respectively. One should note that this comparison was performed under different measurement conditions and thus should only be considered as a general guidline rather than an absolute conclusion.

Download Full Size | PDF

Comparing the slopes of the direct heat and the direct light absorption measurements, it is clear that the responsivity of the SThM probe to direct light absorption is significantly higher ($0.0298\pm 8\text{x}{10}^{-4}$ [nA/mW] as compare to $0.0298\pm 8\text{x}{10}^{-4}$ [nA/mW). Hence, measuring light induced temperature profile of a photonic device using SThM seems to be a challenging task. This conclusion agrees with our previous results [Fig. 2 and Fig. 3] of doped and undoped waveguides in which the optical near field seems to be the dominant mechanism for the generated SThM signal. Yet, we should be careful in exclusively eliminating the significance of light induced heating as a major source of SThM signal. Firstly, our comparison is based on measuring two very different samples which are distributed differently: over the fiber facet in the optical case and along the nanowire in the thermal case. Furthermore they influence on temperature of the junction point differently. The device length may also be an important parameter. For example, our measurements were obtained with a nanowire of 350µm length. Have we dissipated the same power over a nanowire of shorter distance, the responsivity would have been higher. In an extreme case of very short wire, the two mechanisms might become comparable.

Keeping these differences in mind, we believe that it is reasonable to assume that for relatively low absorption waveguides (e.g. doped dielectric waveguides) the dominant mechanism for SThM signal is probably direct light absorption in the probe. On the other hand, for plasmonic waveguides with absorption length in the micron scale, further analysis is needed. A definite conclusion requires further work, beyond the scope of this manuscript.

Following the above mentioned discussion, the rest of the paper is devoted for demonstrating the system capabilities in measuring the optical near field, and for exploring solutions that will allow temperature measurements of nanophotonic devices in the presence of light.

4. Wavelength and frequency response

We now turn into studying the wavelength response of the SThM system. For this task, we use the experimental setup as in Fig. 4(b) and measure the SThM responsivity at four different wavelengths: 2480nm, 1550nm, 1300nm and 980nm. Given the fiber negligible absorption, no profound heat mechanisms are expected in the fiber and the obtained signal is attributed to the optical intensity which is emitted from the facet of the fiber and absorbed by the SThM probe. As expected, one can observe the linear relationship between the obtained signal and the input optical power [Fig. 6]. Furthermore, it is clear that the sensitivity (the slope of the curve) is increasing with the decrease in wavelength. This is because of the higher absorption of light by the metallic SThM probe towards shorter wavelengths.

 

Fig. 6 SThM probe signal vs. incident optical power for four different wavelengths. The measured slopes are 1.36 ± 0.027[nA/mW], 7.31 ± 0.025[nA/mW], 10 ± 0.014[nA/mW], 34 ± 0.012[nA/mW] for the wavelengths of 2480nm, 1550nm, 1300nm and 980nm respectively. Modulation frequency is 704Hz.

Download Full Size | PDF

Next we characterize the frequency response of our SThM. As before, we fix the position of the probe at the center of the optical mode. This time, we change the modulation frequency of the laser (1550nm wavelength) rather than its incident power, while recording the output current obtained from the LIA. The normalized signal as a function of the modulation frequency is shown in Fig. 7. As can be seen, the 3dB modulation frequency is about 1.5 kHz corresponding to response time of about 0.45 msec.

 

Fig. 7 - Frequency response of the SThM probe. The system shows a typical response of a low pass filter wirh a 3dB modulation frequency of ~1.5 kHz.

Download Full Size | PDF

5. Light induce heating measurmet using SThM

The above mentioned experimental results provide a clear indication for the capability to measure the optical near field using SThM probe and shed light of the system performance with respect to wavelength and frequency response. At the same time it becomes clear that the usefulness of the SThM approach for the purpose of measuring the light induced temperature profile of within nanophotonic devices is limited. Following this observation, one may ask the obvious question - is it ever possible to perform the original task of measuring the light induced temperature profile of a nano photonic device using SThM, and if so, under what circumstances?

To answer this question, we first perform a 3D numerical simulation (Comsol multi-physics Ltd) to calculate the electromagnetic field that is absorbed by the metallic probe (for simplicity, the probe is modeled as an Au sphere) which is located above the silicon nano waveguide [see Fig. 8] for two opposite cases of an undoped [Fig. 8(a)] and an extremely highly doped [Fig. 8(b)] silicon nano waveguide. The simulation results show that: 1 – the absorption of light by the metallic probe is nearly identical for both cases, and 2 – even in the case of the extremely highly doped silicon waveguide, the absorption of light by the SThM probe is still far more significant compare to the absorption of light by the silicon nano waveguide. These simulation results further confirm our experimental observations. The reason for the great difficulty in measuring light induced temperature profile of nanophotonic devices is now clear – the optical mode is strongly interacting with the SThM probe and is thus absorbed by its metallic coating. Furthermore, the sharp end of the metallic probe is known to give rise to the mechanism of plasmonic enhanced light absorption. This plasmonic enhanced light absorption in the metallic probe is dominant over the light absorption in the nanophotonic sample, even if the sample is highly doped.

 

Fig. 8 Calculated absorption cross section $\left[\frac{W}{m^3}\right]$ of 1.55µm in plane polarized light for the case of (a) an undoped and (b) an extremely highly doped $\left({\text{~N}}_{\text{h}}=8.5\text{x}{10}^{19}{\text{cm}}^{-3}\right)$ silicon nano waveguide with a metallic probe above it. The gap between the waveguide and the probe is 10nm. The probe is modeled as a 300 nm diameter Au sphere.

Download Full Size | PDF

One possible way to overcome the above mentioned difficulty and measure the temperature distribution of a nano photonic device using SThM is by adding a spacer layer between the nano photonic device and the SThM probe. Such a layer should have a reasonable thermal conductivity, and negligible optical losses. Furthermore, it should be thick enough to prevent the interaction between the optical mode and the SThM probe, while having as small as possible perturbation to the temperature profile of the device. In Fig. 9 the light absorption by the metallic sphere (representing the SThM probe) is plotted against the thickness of the spacer layer (silicon dioxide), see blue line. Additionally, we also plot the temperature of the silicon dioxide layer which is heated due to the light absorption in the doped silicon waveguide $(\kappa =3.7x{10}^{-5}@1.55\mu m\to N_h=5x{10}^{17}c{m}^{-3})$. As can be seen, an oxide layer of ~300nm is sufficient to diminish light absorption in the SThM probe. At the same time, the oxide temperature difference as compared to the ambient temperature varies only by ~10% with respect to the waveguide core upper surface temperature and thus temperature profile can still be measured by the probe and provide a good indication to the original temperature of the device. However, it is important to note that there is a penalty in spatial rsolution. While the temperature profile of the uncoated waveguide is a flat top with dimentions similar to the waveguide width (450 nm in our case), the temperature profile of the coated waveguide is broadened and become a Gaussian like shaped, with a full width half maximum of ~1.5µm.

 

Fig. 9 Comparison between temperature of the oxide layer due to light absorption inside the nano waveguide and light absorption by the SThM probe as a function of the oxide thickness above the nano waveguide. The Optical input power inside the nano waveguide is 10mW.

Download Full Size | PDF

Finally, we discuss the resolution limit while measuring the near field using SThM as compare to the conventional NSOM approach. First, one should differentiate between apertureless and aperture NSOM. For the aperture NSOM, the resolution is limited by the aperture size and the skin depth of the metal. Commonly, the resolution of aperture NSOM is in the order of 50-100 nm for the visible wavelength regime. Apertureless NSOM can provide better spatial resolution. Eventually, the resolution is limited by the scattering cross section. To simplify the discussion, we can assume our probe to hold a sphere-like shape. With such an assumption, the scattering cross section is proportional to $\frac{R^6}{{\lambda }^4}$, where R stands for the radius of the sphere. In contrast, the SThM approach is based on the absorption inside the probe where the absorption cross section is proportional to $\frac{R^3}{\lambda }$. Therefore, as the dimensions of the probe are decreased towards the deep nanoscale, the ultimate resolution of the SThM approach is expected to be better than the resolution of the NSOM. Furthermore, the SThM approach is broadband in its nature, while the NSOM requires specific detector for each wavelength band.

6. Conclusion

In this work we demonstrate numerically and experimentally optical near field characterization of nano photonic devices using an SThM probe. This capability of measuring light by an SThM is attributed to the strong light absorption in the metallic probe, which was shown to be dominant over the mechanism of light induced heating of the dielectric and the semiconductor samples. The system was characterized by several sets of samples of different materials under various wavelengths of operations. All of our measurements indicated that the light absorption by the probe is far more significant than the light induced heat generation in the sample. The experimental results are in good agreement with numerical simulations that were also conducted to support our results. Furthermore, the frequency response of the SThM system was characterized and the 3dB frequency response was found to be ~1.5 kHz.

The approach was shown to be useful for the characterization of variety of samples at different wavelength regimes. The simplicity of the SThM system, together with the broadband wavelength of operation makes this approach an attractive alternative to the more conventional aperture and apertureless NSOM approaches.

Finally, referring to the original role of the SThM probe in the characterization of thermal effects at the nanoscale, we proposed and study an approach for characterizing temperature profile of nanophotonic devices due to the mechanism of light induced sample heating. This is achieved by spatially separating between the optical near field distribution and the SThM probe, while maintaining the temperature profile of the device.