Characterization of electrically active defects at III-N/dielectric interfaces

 
« PreviousUpNext »Contents
Previous: 2.3 Fast measurement with a lock-in amplifier    Top: 2 Experimental aspects    Next: 3 Methodologies for electrical characterization of interface defects in GaN-based MIS-HEMTs

2.4 Setup for opto–electrical characterization

In order to extend the possibilities of the existing electrical characterization setup, a system for optical excitation has been developed. Most of the DLOS experimental setups consist of an open–beam optical bench, shielded by a black enclosure, where a single chip or device can be placed onto a dedicated holder to be characterized. In our case instead, we integrate the optical system to the on–wafer prober. The optical bench is placed outside the probe station and the light is brought to the device under test with a light guide. All the optical components are therefore enclosed individually, to protect the operators from the potentially hazardous radiation when working with wavelengths in the ultraviolet range.

In this section, we describe the components of the optical setup: the light source, the monochromating system and the light guide for the transmission of the optical energy. Furthermore, we evaluate its performance in terms of output light intensity.

Figure 2.9: The spectra of some Xe and Hg discharge lamps. The one used in our setup is the 300 W Xe “ozone free”, labeled with the number 6258 (from [59]).

2.4.1 Light source

The source of light is a high–pressure, short–arc discharge xenon lamp. It has a power of 300 W and emits light in a broad spectrum, almost featureless in the ultraviolet (UV, from 200 nm to 400 nm, or from 3.1 eV to about 6 eV) and visible (VIS, from 400 nm to 700 nm, or from 1.7 eV to 3.1 eV) range, with characteristic Xe lines between 750 nm and 1 µm, as shown in Fig. 2.9. A special coating prevents radiation below 200 nm, which can cause ionization of the oxygen present in the air, producing ozone. This is a safety measure, because breathing ozone can be harmful to health, as it damages the lungs.

The lamp housing (an Oriel Instruments research arc lamp source) consists of a metallic enclosure where the lamp can be mounted on a special socket. The ignitor for the lamp start up is built–in, and a fan with a temperature sensor ensures a proper cooling down of the lamp during operation. A collimating lens system gathers the emitted light and provides an output beam with a diameter of 33 mm. A rear reflector is placed behind the lamp and conveys the light emitted towards the back of the housing into the collimating optics. In this way, the output power is enhanced by 60%. Finally, standard flange mountings allow to match the lamp housing with the next component, leaving the light beam fully enclosed.

Figure 2.10: The working principle of the monochromator (from [60]).

2.4.2 Monochromator

The beam next enters a Cornerstone 260 motorized monochromator, whose function is to transmit selectively the light around the desired wavelength \( \lambda     \), or energy

(2.10) \begin{equation} E = \frac {hc}{\lambda } \approx \frac {1240}{\lambda } \end{equation}

if the wavelength is measured in nm and the energy in eV. The basic principle of operation is shown in Fig. 2.10. The input slit lets the light in onto the grating, which diffracts the beam by reflecting the components at various wavelength with a different angle. The grating can rotate around its axis, thus selecting the central wavelength that is conveyed by the focusing mirror into the exit slit. A mechanical shutter can be used in order block the light path. The slits can be manually adjusted by micrometer–driven screws from a minimum width of 4 µm to a maximum of 3 mm, and from 3 mm to 12 mm in height. The width of the exit slit determines the bandpass, i.e., the range in wavelength of the output light: the thinner the slit width, the narrower the bandpass. The slit height instead has an impact on the output light intensity.

The gratings consist of an aluminum surface with regular, parallel grooves. The grating equation can be derived considering Fig. 2.11a: we have constructive interference between the reflected rays A\( ^1 \) and B\( ^1 \) if

(2.11) \begin{equation} \mr {a} \left ( \sin {\alpha } + \sin {\beta } \right ) = \mr {m} \lambda   \end{equation}

where \( \mr {m} \) is an integer and represents the order of diffraction. The most intense reflection corresponds to the first order \( \mr {m} = 1 \). This basic equation holds true not only for two parallel grooves but for the whole grating. As a consequence, each wavelength is reflected with a certain angle \( \theta   \) for \( \mr {m} = 1 \), and at multiples of \( \theta              \) at higher orders. If a white light beam is the input of the monochromator, it is clear that in certain situations we might obtain the main first order contribution from the wavelength \( \lambda   \), but also an additional contribution from the second order of the wavelength \( \lambda /2 \) at the same angle. This problem is illustrated in Fig. 2.11b, and it is solved by using a cut–on filter when the desired output is at a multiple of the minimum wavelength. In our case the minimum achievable wavelength is around 200 nm, therefore a cut–on filter with cut–off wavelength of 400 nm is used when the output is set above 400 nm in order to suppress second order reflections of light at 200 nm, and similarly a cut–on filter at 800 nm is used to eliminate the third order contributions.

(image)

Figure 2.11: (a) Illustration of the light diffraction by the grating’s surface, from which the grating equation is derived. (b) Diffraction of polychromatic light from the grating: different orders of different wavelengths can overlap (adapted from [60]).

The Cornerstone 260 monochromator allows to mount two gratings on a rotating stage, so that the user can select the desired one. The characteristics of the gratings are summarized in Table 2.1. The lines per millimeter indicate the density of grooves on the grating surface: the larger this value, the higher the output intensity and the narrower the wavelength range that can be achieved. The efficiency of the grating is not constant throughout its wavelength output range: the one that maximizes the efficiency is called the blaze wavelength. Finally, the reciprocal dispersion determines the final bandpass as a function of the exit slit width.

LR grating HR grating
Lines per millimeter in /mm 150 1200
Blaze wavelength in nm 300 350
First order range in nm 190 to 800 200 to 1400
Peak efficiency 70% 80%
Reciprocal dispersion in nm/mm 25.5 6.7

Table 2.1: Characteristics of the two gratings in use with the Cornerstone 260 monochromator.

We use the first grating to maximize the bandpass (LR or low resolution grating), because in this way we maximize also the light intensity. In this case, for the maximum slit width of 3 mm the bandpass is 25.5 nm/mm \( \times   \) 3 mm \( = \) 160 nm. The second grating (HR or high resolution grating) is dedicated to spectroscopies, therefore the aim is to achieve the best resolution in wavelength. If we keep the slit width at its maximum to improve the light intensity, the bandpass is 6.7 nm/mm \( \times   \) 3 mm \( = \) 20 nm, and it can be decreased further by making the slit narrower. Although the theoretical maximum resolution is 0.02 nm, the real value is limited to 0.1 nm because of non–ideal behavior of the optics, for example due to aberrations.

Except for the micrometer driven slits, all other functionalities of the monochromator can be accessed remotely via GPIB. Dedicated drivers and a control software have been written in LabView. Available commands allow to choose the grating in use, set the central wavelength and manage the shutter opening and closing time. It is also possible to perform the same operations within the main measuring program described in Section 2.2, using a dedicated syntax in the input command file.

2.4.3 Light transmission

The light beam at the monochromator’s output is ready to be brought to the device under test inside the on–wafer prober. We use a 2 m–long Newport liquid light guide for efficient transmission of light. It consists of a plastic tube, covered by a PVC jacket. The inner part is filled with a transparent, anaerobic, non–toxic fluid which transports optical energy by total reflection. The active core is 5 mm in diameter, sealed at both ends with fused silica glass windows. The whole volume is uniformly filled with light during operation, in contrast with fiber optics which are formed by many single, smaller tubes, thus creating a pattern of active and dead spots. The transmittance range of the light guide is dedicated to wavelengths in the UV and VIS range, as shown in Fig. 2.12. The efficiency is very good, exceeding 80% around 500 nm.

Figure 2.12: Transmission efficiency of the liquid light guide used in our setup.

Since the working principle is total reflection, the light guide has a defined acceptance angle. This means that light rays with larger angle than the critical \( \theta                         \) are not transmitted but absorbed, thus heating up the light guide and eventually wearing out the light conductive material. The sine of such angle is called numerical aperture (NA) and it is a characteristic of the light guide, determined by its geometry and the refraction index of the building materials [61]. The NA can be calculated for any optical component, in the way illustrated in Fig. 2.13a. In the case of a lens, the important parameters are its diameter \( D \) and its focal length \( f \). The NA is calculated with the approximation for small angles, resulting in \( D/2f \). This number quantifies the light gathering ability of the lens, or fiber: the larger the NA, the larger the acceptance angle.

(image)

Figure 2.13: (a) Illustration of the numerical aperture (NA) definition. (b) Coupling the light guide to the monochromator requires a proper choice of focusing optics (adapted from [60]).

For example, in order to couple the light guide to the monochromator’s exit slit, we must use a focusing assembly that matches the NAs of the two parts. The monochromator and the light guide have NAs of 0.13 and 0.43, i.e., the monochromator’s output divergence angle is \( 2\theta = \) 15 degrees, and the light guide has an acceptance angle of \( 2\theta = \) 98 degrees. We can efficiently focus the light beam using a bi–convex lens of diameter \( D= \) 25.4 mm and focal length \( f= \) 32 mm, as shown in Fig. 2.13b. This works because the lens’ NA is 0.385, thus resulting in an angle of 90 degrees. In this way, the light from the monochromator can be fully gathered by the lens, and then focused into a volume which is contained by the acceptance cone of the light guide, so that the incidence angle always results in total internal reflection.

The light guide is flexible and it can be introduced inside the probe station. It can be attached to an adjustable mounting, anchored to the bench with a magnetic base, which can be positioned and fixed above the device under test in order to perform experiments including optical excitation.

2.4.4 Light intensity

The intensity of the light provided by the light guide is not constant at all wavelengths. It is important to evaluate the relative intensity of light over the accessible energy range in order to correctly interpret the response of the device under test when it is exposed to optical excitation at various wavelengths.

In the first place, the lamp radiates according to the emission spectrum, shown in Fig. 2.9. Furthermore, every optical component transmits different wavelengths with variable efficiency. The optical power at a certain wavelength is the product of the lamp emission spectrum with the transmission or reflection spectra of all components in the optical path: condenser lens, grating, cut–on filter, focusing lens and the light guide. Finally, the power of incident light onto the device is found by integrating the power at a single wavelength over the range of interest, which is determined by the bandpass of the monochromator.

Figure 2.14: Spectral response of the calibrated photodiode (data from [62]).

While it is difficult to give an accurate estimate of the optical power based on calculations, it is possible to measure it with a calibrated photodiode. Photodiodes are sensors that produce a current directly proportional to the intensity and wavelength of the light incident on their active area. They are characterized by a specific spectral response, called responsivity, defined as the amount of current produced per unit power of incident light as a function of the wavelength. We use a Thorlabs SM1PD2A photodiode, whose responsivity is shown in Fig. 2.14.

In this way, we can evaluate the power of our optical setup as a function of the wavelength (or energy) performing a measurement with the photodiode. Recording the current generated by the light, the photocurrent \( \mr {I_{photo}} \), and knowing the photodiode’s responsivity R, we can calculate the power \( P_\mr {light} \) at a single wavelength as

(2.12) \begin{equation} P_\mr {light} = \frac {\mr {I_{photo}}}{\mr {R}}.                           \end{equation}

If the sensor’s active area is illuminated by the whole light beam without losses, we can calculate the number of photons arriving on the surface, or the photon flux \( \Phi _\mr {ph} \) as

(2.13) \begin{equation} \Phi _\mr {ph} = \frac {P_\mr {light}}{q \:   \mr {area} \:   h\nu } \end{equation}

where \( q \) is the elementary charge, the area is that of the photodiode’s active surface, and \( h\nu = hc/\lambda               \) is the photon energy. The result is shown in Fig. 2.15: our optical setup is capable of generating about 1.5 mW of optical power on average, providing up to \( 2 \times 10^{15} \) photons per unit area and per unit time.

(image)

(image)

Figure 2.15: Incident optical power (a and b) and photon flux measurement (c and d) with the calibrated photodiode as a function of the photon wavelength and energy.

These results will be used to normalize the device’s electrical response during exposure to different wavelengths. Nevertheless, since the area of the device is much smaller than that of the illuminated region by the light beam, the effective power (and number of photons) incident on the structure under test will be smaller than what has been calculated here, but still being proportional to that. For this reason, any quantity evaluated as a function of the light energy will include a proportionality factor, as we will see in Chapter 6.

« PreviousUpNext »Contents
Previous: 2.3 Fast measurement with a lock-in amplifier    Top: 2 Experimental aspects    Next: 3 Methodologies for electrical characterization of interface defects in GaN-based MIS-HEMTs