GB2390152A - Waveguide absorption measurement - Google Patents

Abstract

There is provided a method of measuring the loss or absorption coefficient of a semiconductor waveguide. The method is suitable for waveguide loss measurement in passive waveguides, where coupling efficiency considerations become insignificant, and especially for the case of low loss short length waveguide components. The input light signal is divided into at least two output channels (C1, C2, C3, C4 and reference) by a multimode interferometer (MMl) 1. A waveguide is coupled to each of the output channels, at least two of the waveguides having different lengths. The output power from each of the waveguides is measured at the output of each waveguide, preferably using a photodetector. Preferably, at least one of the waveguides is cut back, leaving a reference channel uncut at the same length. The input power is then adjusted so that the output power in the reference channel is the same as the first measurement. The new output power levels in the cut-back waveguides are measured again. The loss or absorption coefficient is then calculated from the differences between measured values of output power before and after being cut-back.

Description

WAVEGUIDE ABSORPTION MEASUREMENT

Field of the Invention

The present invention concerns a method of characterizing active or passive 5 semiconductor waveguide devices, and particularly a method of measuring the absorption (or loss) coefficient of such a device.

Background to the Invention

Semiconductor waveguides are used to route optical signals on a 10 semiconductor chip, which provides the basis for numerous integrated optic devices (such as transmitters, receivers, semiconductor optical amplifiers, filters, switches, and so on) suitable for application to optical communications and optical signal processing. From the measurement point of view, accurate and fast characterization of waveguide parameters is highly desirable. The toughest challenge comes from 15 the loss measurements. The loss measurement is also the basis of other parametric characterization of passive components.

Waveguide loss includes the coupling loss and intemal loss. The coupling loss occurs at the interface between the waveguide and other optic components, for example an optical fiber or a micro-lens. The intemal loss can be classified as either 20 absorption, scattering or leakage. Absorption arises from imperfect material transparency. Scattering results from transfer of optical energy from guided mode to radiation mode caused by imperfect refractive index distribution. The transfer of light causes leakage from the guided layer to the substrate layer.

The simplest method for evaluating waveguide loss is the measurement of 25 the attenuation using transmission experiments. The ratio of output power to input power yields the waveguide loss (including the coupling loss and the intemal loss).

Most of time, the coupling loss and the intemal loss need to be separated from each other for the purpose of characterization of waveguide properties. The standard approach used to distinguish coupling and intemal losses in waveguide samples is 30 to measure the loss as a function of the length of waveguide (cut-back method), which is shown in Figure 1. The major difficulty with this technique is input coupling reproducibility, which is typically 0.2 dB. The intemal loss of the waveguide ranges from 0.2 dB/cm to 5 dB/cm, which depends on the material and structures. In most modem optical components, the length of the waveguide is less than 1 cm, and 35 hence the intemal loss is very small and insignificant with respect to the coupling loss. Thus the effect of coupling loss reproducibility becomes relatively serious. As a result, the intemal loss coefficient a cannot be precisely determined.

Other methods of measurement of waveguide loss include optical scattering measurement as well as activation of the active laser waveguide. However, both of them suffer from low accuracy, which makes them only applicable for waveguides with long length (several centimeters) or with special structures. Obviously they are 5 not suitable for small and short waveguide measurement.

Semiconductor photodetectors are used extensively in optical communications for detection of optical signals. First, photodetectors are used to convert optical signals to electrical signals for further processing. Second, photodetectors are used as monitoring devices, providing information on the signal 10 strength and drift, so that suitable adjustment can be made to the signal for further propagation. From the measurement point of view, photodetector parameter extraction has always posed a challenge. The main difficulty, especially for waveguide detectors, is to determine the exact amount of power incident to the detector input 15 surface. In most cases, the coupling loss, propagation loss arising from the absorption and scattering loss, could not be determined easily. Certain forms of assumption or approximation are normally applied and the results obtained are based strongly on the reliability of the assumptions made.

Two main parameters for photodetectors, the absorption coefficient, which is 20 the amount of power absorbed (in dB) per unit length, and responsivity, which is the ratio of the detector output current versus the input power, can be deduced only with the knowledge of the input power. Thus, it is desired to find a good method to eliminate the uncertainty in the input power to the waveguide detectors, which would improve the accuracy of the obtained results tremendously.

Summary of the Invention

According to the present invention, a method of measuring the loss or absorption coefficient of a semiconductor waveguide comprises the steps of: providing an input light signal to a multimode interferometer (MMI), the MMI 30 having at least two output channels; coupling a waveguide to each of the output channels, at least two of the waveguides having different lengths; measuring the output power from each of the waveguides; and calculating the loss or absorption coefficient.

35 In one embodiment, the method further includes the steps of; cutting back at least one of the waveguides and leaving a reference waveguide the same length;

adjusting the input power so that the output power of the reference waveguide is the same as in the first measurement; and measuring the output power of the cut back waveguides.

In this method, the coupling efficiency considerations would become 5 insignificant, especially for the case of low loss short length waveguide components, where the traditional methods of measurement cannot support accurate characterization. In this embodiment, preferably a plurality of waveguides are used in addition to the reference waveguide. This improves the accuracy of the method.

10 In this embodiment, preferably the waveguides are passive waveguides.

According to another embodiment, the power is split equally by the MMI into each output channel. This enables the removal of the input power term in the calculation, providing better accuracy in the obtained results.

In this embodiment, preferably multiple measurements are taken using 15 waveguides of different lengths, and the mean value of the loss or absorption coefficient is calculated.

In this embodiment, preferably a large area photodetector is used to measure the output power from each of the waveguides. This provides a coupling efficiency of substantially 100% at the output ends.

20 In this embodiment, preferably the waveguides are active waveguide devices, particularly photodetectors.

Brief Description of the Drawings

Examples of the present invention will now be described in detail with 2s reference to the accompanying drawings, in which: Figure 1 illustrates a prior art cutback method;

Figure 2 illustrates step 1 of a method in accordance with a first embodiment of the present invention; Figure 3 illustrates step 2 of the method of the first embodiment; 30 Figure 4 illustrates step 3 of the method of the first embodiment; Figure 5 illustrates the derivation of internal loss coefficient a in the first embodiment; and Figure 6 illustrates a second embodiment of the present invention.

35 Detailed Description

The present invention makes use of a multi-mode interferometer (MMI) 1 as a splitter in the measurement.

Figures 2 to 5 illustrate a first embodiment suitable for waveguide loss measurement of passive waveguides, in which the coupling efficiency considerations become insignificant, especially for the case of low loss short length waveguide components. The input light signal is divided into five channels C1, C2, 5 C3, C4 and ref. by the MMI 1. The channel spacing is about 5 mm for ease of detection at the output of each waveguide using a photodetector. The photodetector, which provides a larger detection area, ensures that the coupling efficiency between the waveguide and detector is 100%. Note that the length of each different channel C1, C2, C3, C4 and ref. is different as shown in Figure 2.

10 Figure 2 illustrates the first step of the method, in which the MMI structure is loaded to the test station and light is coupled into the MMI 1. Next, the output power Pi', P2', P3', P4' Pa,' of each channel is measured (in units of dam) using the photodetector. Figure 3 ililustrates the second step, in which the MMI sample is unloaded 15 and the waveguides are cut to the same length, the cutoff lengths of channels C1, C2, C3, and C4 being L1, L2, L3 and L4, respectively. The reference channel waveguide is not cut.

Figure 4 illustrates the third step, in which the MMI is reloaded and the output powers P., P2, P3 P4, Prep are measured again. As the coupling of the input 20 channel is not the same as that in previous measurement, it is necessary to increase or decrease the input power until the output power of channel ref. Pelf is equal to previously measured Ptrf' which indicates that the output power distribution of the different waveguide channels C1, C2, C3, C4 is the same as in the previous measurement. The output power of different channels C1, C2, C3, C4 are recorded 25 as Pi, P2, P3, and P4, respectively, as shown in Figure 4.

Pi (in units of dam) minus P. represents the internal loss of waveguide with length of L1; P2 minus P2 represents the internal loss of waveguide with length of L2, and so on. The internal loss of waveguides with different lengths can be calculated. Figure 5 shows the curve of loss Pa against waveguide length L. The 30 internal loss coefficient a can then be easily obtained by fitting the curve and calculating the slope.

p=P-P(dB) (1) 35 a = ddPL (dB I cm) (2)

The advantage of this method is the coupling independency. Although the basic idea comes from the traditional cut-back method, the accuracy is considerably improved to be limited only by the accuracy of the measurement equipment. This 5 method is especially useful for waveguide samples with lengths shorter than 5 mm, in which it is almost impossible to measure using other methods, such as the traditional cut-back, optical scattering and activation of the active laser waveguide methods. Figure 6 illustrates a second embodiment of the invention suitable for 10 characterizing waveguide photodetectors. This method also makes use of a MMI 1 as a splitter in the measurement. Using a 3dB MMI 1, the input light signal is divided into two channels, each of which essentially contains 50% of the input signal power. This is done so as to ensure the amount of light incident to both the output detectors 2, 3 is identical. As shown in Figure 6, at the output of the MMI 1, two 15 detectors 2, 3 of different lengths are placed. The difference in length for the detectors 2, 3 will provide data points for the calmiation of detector parameters. In order to obtain data with high accuracy, the detector lengths have to be appropriate.

The lengths should not be too long as both detectors 2, 3 will fully absorb the light, and if the lengths are too short very little absorption will occur. Ideally, it is best for 20 the shorter detector 3 to absorb a portion of the light, and the longer detector 2 to almost fully absorb the light. As the absorption coefficient a is the unknown, it is hard to determine only one set of appropriate lengths for the detectors 2, 3, thus a variation of detector lengths needs to be designed and the accurate results obtained by statistical means.

25 After fabrication, the waveguide detector structures would be cleaved and measured. We only need to ensure the MMI 1 splits the power equally to the two output detectors 2, 3 and sufficient power is provided. The output powers Pi, P2 at the back facet of the detectors 2, 3 are then measured, and the absorption coefficient a could be expressed as follow: P' = P;n (ELI (dBm) P2 = Pin - aL2 (dBm) (4) where: as P',2 = Output power after waveguide photodetector

P,n = Power propagating from output port of MMI into input of waveguide photodetector (dam) a = Absorption coefficient of detector L.,2 = Length of varied detector 5 Equation (3) minus equation (4) gives A, - P2 = aL2aL, () We see that the input power term (P'n) is eliminated from the equation. The 10 value for the absorption coefficient a could then be calculated by solving the equation using numerical means: a= ' 2 (dBlem) (6) L2 - L,

15 The absorption coefficient value a for the same set of detector lengths L,, L2 needs to be taken a few times so that an average value with standard deviation can be computed. Different MMI detector sets with different length variations also need to be analyzed to obtain an accurate value for absorption ooeffcient a.

Upon obtaining the value for absorption coefficient a, we can further extract 20 the responsivity value of the photodetectors 2, 3. From equation (3), we could easily obtain the value for Pun, since both a L, and Pa are of known value.

The responsivity R can be easily calculated by dividing the measured output current lo'' from the photodetector by the input power.

25 R It's' (7) P'n

Claims (8)

1. A method of measuring the loss or absorption coefficient of a semiconductor waveguide, comprising the steps of: 5 providing an input light signal to a multimode interferometer (MMI), the MMI having at least two output channels; coupling a waveguide to each of the output channels, at least two of the waveguides having different lengths; measuring the output power from each of the waveguides; and, 10 calculating the loss or absorption coefficient.

2. A method according to claim 1, further comprising the steps of: cutting back at least one of the waveguides and leaving a reference waveguide the same length; 15 adjusting the input power so that the output power of the reference waveguide is the same as in the first measurement; and, measuring the output power of the cut back waveguides.

3. A method according to claim 1 or 2, in which a plurality of waveguides are 20 used in addition to the reference waveguide.

4. A method according to claim 3, in which said plurality of waveguides are passive waveguides.

25

5. A method according to claim 1, in which the power is split equally by the MMI into each output channel.

6. A method according to claim 5, in which multiple measurements are taken using waveguides of different lengths, and the mean value of the loss or absorption 30 coefficient is calculated.

7. A method according to claim 5 or 6, in which a large area photodetector is used to measure the output power from each of the waveguides.

35

8. A method according to any of claims 5 to 7, in which the waveguides are active waveguide devices.