GB2386414A

GB2386414A - Waveguide propagation loss measurement method

Info

Publication number: GB2386414A
Application number: GB0205427A
Authority: GB
Inventors: Yuen Chuen Chan; Yee Loy Lam; Seng Lee Ng
Original assignee: DenseLight Semiconductors Pte Ltd
Current assignee: DenseLight Semiconductors Pte Ltd
Priority date: 2002-03-14
Filing date: 2002-03-14
Publication date: 2003-09-17
Also published as: GB0205427D0

Abstract

A method of measuring the waveguide loss coefficient of a semiconductor waveguide 1 comprises the steps of injecting light into the waveguide and measuring the lost light by measurement of the photocurrent generated. The method is thus independent of coupling efficiency, as no measurement of the input or output light is required and no cutting of the waveguide is required. The photocurrent can be detected externally from the surface of the waveguide by providing p and n type contacts 3, 4 on the respective surfaces of the waveguide. The waveguide may be divided into N sections each having an individual contact 3, with a single contact 4 provided on the other surface of the waveguide. The generated photocurrents I1 - IN can then be measured sequentially for each section.

Description

- 238641 4

WAVEGUIDE PROPAGATION LOSS MEASUREMENT METHOD

Field of the Invention

Confinement of light in planar non-fibre based optoelectronic devices is 5 achieved through waveguide structures, where a region of higher refractive index than its surroundings acts as a light guide. Such waveguide structures have been demonstrated widely on III-V compound semiconductors. Almost all active devices, like laser diodes, modulators, optical amplifiers and detectors, as well as passive devices, like Y- branches, couplers, tapers, and arrayed waveguide gratings are 10 based upon this waveguide structure. The present invention concerns a non-

invasive method for determining the propagation losses of a semiconductor waveguide structure.

Background to the Invention

15 Devices based upon the waveguide structure, as shown in Figure 1, have many loss sources, which constitute the waveguide loss. The sources include absorption, scattering, and leakage. Absorption is due to the imperfect material transparency, while scattering and leakage result in shifting the optical energy from the guided modes to radiation modes. Scattering occurs from imperfect refractive 20 index distribution, mainly due to interface roughness, whereas leakage takes place even for structures with ideally smooth sidewalls. The exponential attenuation of the optical power in the guided mode is described by the waveguide loss coefficient (a).

The most conventional and simplest method for characterizing waveguide loss is by the transmission measurement. The ratio of output to input optical powers 25 as illustrated in Figure 2, yield the overall losses, which includes the coupling losses (,0' and p2) at the input and output ends of the waveguide, and the waveguide loss (or). The coupling loss,8' refers to the coupling of the input power from the light source into the waveguide, while p2 refers to the coupling of power out from the waveguide. The equation is as follows: P ut = {12 expand (1) where L is the length of waveguide and P'x= 0) p2 = Phi' (2)

are the coupling losses. Here, P'(x) denotes the light that is propagating along the waveguide at the distance x from the input facet. Here, there is a need to distinguish between the coupling and waveguide losses so that the waveguide loss can be determined accurately.

5 In this prior art method, the loss is measured as a function of the waveguide

length L, and this technique is usually called the cutback method. The purpose of measuring the total loss variation with the waveguide length is to separate the effect of the coupling losses,B, and p2 (which are independent of the waveguide length) from the waveguide loss a. However, the major difficulty with this method is the 10 reproducibility of the input and output coupling conditions, where the coupling alignment has to be done again and again as the length of waveguide is varied for the data point collection. Consequently, this limits the overall accuracy of the measurement, particularly when the variability of the coupling losses,8' and p2 leads to a large scattering of the data points obtained from the measurement.

Summary of the Invention

According to a first aspect of the present invention, a method of measuring the waveguide loss coefficient of a semiconductor waveguide comprises the steps of: 20 injecting light into the waveguide; and measuring the lost light by measurement of the photocurrent generated.

The method of the present invention is independent of the coupling efficiency, as coupling realignment is not necessary for this technique, as no measurement of the input or output light is required and no cutting of the waveguide 25 is required. The photocurrent can be detected externally from the surface of the waveguide. Preferably' the injected light has an energy which is lower than the band gap energy of a core material of the waveguide. Although it is possible to employ the method with light of a higher energy than the absorption band gap, as some 30 absorption will be caused, it is preferable to use light of an energy lower than the band gap, as waveguides are normally utilized in the transparency spectrum.

Preferably, p and n contacts are provided on the surface of the waveguide for measurement of the photocurrent.

Preferably, a plurality of sets of p and n contacts are provided on the surface 35 of the waveguide. single n contact may be provided on the n side of the waveguide, with individual p contacts spaced along the p side, or vice versa.

Preferably, the sets of contacts are spaced evenly along the length of the waveguide. Preferably, the method includes making measurements of the photocurrent at each of the sets of contacts non-simultaneously.

S Preferably, the photocurrent is passed through a resistor and the voltage across the resistor is measured.

Preferably, the waveguide is a III-V semiconductor waveguide.

According to a second aspect of the present invention, a semiconductor waveguide includes p and n contacts on the p and n surfaces of the waveguide 10 respectively, the p and n contacts being arranged along the length of the waveguide, to allow measurement of the photocurrent at intervals along the length of the waveguide. Preferably, the sets of p and n contacts are evenly spaced along the length of the waveguide.

15 Preferably, the waveguide is a III-V semiconductor waveguide.

Preferably, the waveguide includes means to measure the photocurrent generated at the p and n contacts as a result of the waveguide loss.

Brief Description of the Drawings

20 Examples of the present invention will now be described with reference to the accompanying drawings in which: Figure 1 shows a semiconductor waveguide; Figure 2 shows a cross-section of the waveguide; Figure 3 illustrates the method and the waveguide in accordance with the 25 present invention; Figure 4 shows a plot of 21n(1N/1) against (N-1) (x + y); and Figure 5 shows a plot of x against N and L against N for different ratios of (INII1)

Figures 1 and 2 show a semiconductor waveguide 1 having an active region 30 2. Input power Pin for the light source is input at one end of the waveguide 1, is confined in the active region 2, and is output POU' at the other end of the waveguide 1. The variation of the power with distance x along the waveguide can be expressed by the equation: P (x) = POexp (-ax) 35 Wherein PO is the power at x=O, and PO = p' Pin.

The output power PoU, at the end of the wavelength of length L is given by the equation:

Pout = POp2exp (-aL) In the prior art cut-back method, a number of measurements of POU! are

made, cutting back the length L of the waveguide with each measurement.

Figure 3 shows the new proposed setup for waveguide loss measurement in 5 accordance with the present invention. In this technique, the waveguide length L is divided into N sections. Each section of the short waveguide 1 has a p- type contact 3 on the p-side side and a single e-type contact 4 is provided on the e-side, wherein each p-contact 3 has a length of x. Adjacent p contacts 3 are spaced by a distance y, so that the total length of each section is x y. Within each section, there will be 10 waveguide loss as a light wave traverses through the active layer 2 of the waveguide 1. This is due to the fact that optical absorption occurs as the light wave propagates through the waveguide, in addition to the possible scattering and leakage losses. For implementation of this example in measuring substantially transparent waveguides, the input light source energy bandedge should be lower 15 than the absorption bandedge energy of the core material of the waveguide under test, i.e. the wavelength should be longer than the bandgap wavelength of the core material. As light propagates through the waveguide and absorption occurs, photocurrents will be generated at each of the metal contacts 3, 4 (I', It,...., /N) The 20 current (I', 12...., /N) generated can be measured by means of an ammeter A or by conversion to a voltage signal through a resistor R. As the optical power propagates through the waveguide 1, the optical power will decrease and the variation can be expressed by the exponential power attenuation equations: 25 For section 1: P. = PO exp(- a[x + y]) (3) For section 2: P2 = PI exp(-Ct[x + y]) (4) 30 For section N: PN=PN exp(-a[x+y]) (5) Here, PO is the optical power coupled into the waveguide, while P. refers to the optical power remaining after travailing through the ith section.

As a result, with nearly complete absorption of the light within the length of x, 35 the current flows through each metal contact can be expressed as:

For section1: I,2= R {l-exp(-o![x+y])} (6) For sections: I22 = R{l-exp(z[x+y])} (7) For section N: I,,, 2 - R-l {l-exp(-[x+y])} (8) where I 2 = ptR from the basic ohm's law.

By substituting equation (3) - (5) into (6) - (a), we get: For section 1: II 2 = R {1 - exp(-a[x + y])} (9) 10 For section 2: I 2 = Po exp(-cr[x+y]) {l-exp(-z[x+y])} (10) For section N: IN2 = R exp( (N-l)cz[x+y]){l-exp(cx[x+y])}(11) By taking ratios of /,, 12, /N with l,, equations (9)-(11) can be generalized as follows: iN = exp(-(N-l)c[xy]' (12) 20 which suggests that the effect of coupling loss has been eliminated from the measurements. By taking natural log on both sides, equation (12) becomes: I 2 In N2 = -(N - ba,rx y] (13) From equation (13), it shows that the waveguide coefficient (a) can be measured by plotting the straight line of lnI2/I,2 versus (N - 1)[x + y] for N = 1,

2, N data points, as shown in Figure 4. The waveguide loss coefficient (a) can be obtained from the slope of the straight line. Since the straight line passes through the origin, one needs to carry out a minimum of two measurements IN and l' to determine the waveguide loss coefficient a.

5 Figure 4 shows that the straight line joining up the data points will always pass through the origin, implying that the coupling loss coefficient has been isolated from the experimental measurements. Thus, this shows that this novel technique is independent of coupling reproducibility, which will enhance the accuracy and reproducibility of the overall measurements. In additional, this method offers a 10 quicker measurement, as compared to the cutback method, since the time consuming coupling re-alignment for each varied waveguide length is avoided.

Consequently, as the measurement takes a shorter time, it is also insensitive to the wavelength and power stability of the input source used.

It should be noted that the measurement should be carried out one section at 15 a time. This is because the fumed-on section nearer to the input facet will absorb a substantial amount of the propagating light, leaving almost no remaining light to propagate down the waveguide 1, and hence a section lower down will not be able to register any signal if it is fumed on at the same time as the first.

For ease of fabricating the alternating metal contact pads onto the 20 waveguide structure, the gap y could be set to be 20,um. For substantially waveguides, a is usually of the order of less than 1 /cm. For any photocurrent (/,, 12, IN) to be detectable at any part of the metal pad 3, the current flow must exceed the dark current when no light is injected into the waveguide. The dark current is usually in the range of 1 PA. Hence, the measurable current for the last Nth section 25 /N should be larger than 1 IJA. Assuming that the a is uniform across the whole waveguide length and is equal to 1 /cm, for a clear distinction between each individual current flow from each metal pad, it is necessary to have a reasonable margin between photocurrents /, and /N.

For example,

30 For IN/I, = 1/5, 3.2 = (N - 1)(x + 0.002) (14) For IN/11 = 1/10, 4.6 = (N - 1)(x + 0.002) (15) For IN/I] = 1/100, 9.2 = (N - 1)(x + 0.002). (16) Figure 5 shows that as IN/I, increases, the length of the pad x decreases, which in turn decreases the overall length of the waveguide L. For IN/11 1/5, the 35 length of the waveguide can be as long as 3.5 cm, which is typical for waveguide loss measurements.

Claims

Claims

1. A method of measuring the waveguide loss coefficient of a semiconductor waveguide comprising the steps of: injecting light into the waveguide; and 5 measuring the lost light by measurement of the photocurrent generated.

2. A method according to claim 1, wherein the photocurrent is detected externally from the surface of the waveguide.

10 3. A method according to claim 1 or 2, wherein the injected light has an energy which is lower than the band gap energy of a core material of the waveguide.

4. A method according to any one of the preceding claims, wherein p and n contacts are provided on the surface of the waveguide for measurement of the 1 5 photocurrent.

5. A method according to any one of the preceding claims, wherein a plurality of sets of p and n contacts are provided on the surface of the waveguide.

20 6. A method according to any one of the preceding claims, wherein a single n contact is provided on the n side of the waveguide, with individual p contacts spaced along the p side, or vice versa.

7. A method according to claim 5 or 6, wherein the sets of contacts are spaced 25 evenly along the length of the waveguide.

8. A method according to any one of claims 5 to 7 including the step of making measurements of the photocurrent at each of the sets of contacts non-

simultaneously. 9. A method according to any one of the preceding claims, wherein the photocurrent is passed through a resistor and the voltage across the resistor is measured. 35 10. A method according to any one of the preceding claims, wherein the waveguide is a III-V semiconductor waveguide.

11. A semiconductor waveguide including p and n contacts on the p and n surfaces of the waveguide respectively, the p and n contacts being arranged along the length of the waveguide, to allow measurement of the photocurrent at intervals along the length of the waveguide.

12. A semiconductor waveguide according to claim 11, comprising sets of p and n contacts which are evenly spaced along the length of the waveguide.

13. A semiconductor waveguide according to claim 11 or 12' comprising a single 10 n contact on the n side of the waveguide, with discrete p contacts spaced along the p side, or vice versa.

14. A III-V semiconductor waveguide according to any one of claims 11 to 13.

15 15. A semiconductor waveguide according to any one of claims 11 to 14, including means to measure the photocurrent generated at the p and n contacts as a result of the waveguide loss when light is injected into the waveguide.