Characterization of External Cavity Diode Laser

Characterization of External Cavity Diode Laser C. Bean,1 A. Gardner,1 C. Noldner,1 and M. Tennakoon1 1

Augustana College, Sioux Falls, SD 57197, USA (Dated: March 6, 2015)

A series of experiments are performed to characterize the light emitted from an external cavity diode laser. We measure the threshold current to be 35.005 ± 0.00074899 mA. Next we use interferometry to make a precise measurement of the central wavelength of the diode laser, which is found to be 780.22 ± 0.02489 nm. The spectral bandwidth of the laser is measured to be 1.49891 ± 0.04897 nm, and by comparing the data with the data sheet we determine that the diode is free-running. We also analyze the interference fringes produced by scanning the wavelength of the diode laser using the Scan Control 110 and determine the tuning range for the laser frequency to be 4.14 ± 0.5GHz. PACS numbers:

I.

INTRODUCTION

Diode lasers have become important instruments in atomic physics because of their reliability in producing narrow-band light (< 1 MHz) and are less expensive than tunable dye lasers. Besides their applications in modern technology such as optical data storage and laser pointers, diode lasers are popular in spectroscopy and microscopy. A tunable diode laser typically consists of tiny Gallium-Arsenide semiconductors that generate coherent light in a very small package. Lasing occurs as a result of the difference in energy levels between the conduction and valence band electrons in these semiconductors. In a typical configuration, the diode has a thin layer of doped material on the surface of an n-type crystal substrate wafer. The doping creates an n-type region and p-type region that results in a p-n junction (or diode). When the diode is forward-biased, holes from the p-type region are injected into the n-type region, which forms an active region where electrons and holes recombine resulting in light emission. The laser diode also contains an optical cavity, usually created by cleaving the semiconductor crystal to form perfectly parallel ends on either side (Fabry-Perot resonator). This is where stimulated emission occurs, which is when an incoming photon triggers the recombination of an electron and hole, with emission of a second photon with the same frequency. Within the laser cavity is a thin waveguide that is terminated on each end by a mirror. Photons emitted into the waveguide are reflected multiple times at these ends, and each time, they are amplified by stimulated emission. The creation of multiple photons traveling in the same direction as the incoming photon with the same energy and phase results in a coherent beam. In laser spectroscopy, we need to determine the energy levels of a sample as accurately as possible; therefore, it is necessary to minimize the width of the spectral output of the laser. This can be done by feeding a small part of the laser output back into the waveguide. In this experiment, we use an External Cavity Diode Laser (ECDL), which uses an external diffraction grating, which can be

tilted to tune the wavelength of the laser over the supported frequency range. The wavelength can be adjusted by tuning the current and temperature of the diode, but this is only a coarse adjustment as line widths can exceed 1MHz. Also, mode hops can occur; this is a phenomenon where the laser exhibits sudden jumps of optical frequency caused by transitions between different modes of its resonator [1], and is characterized by a leap over a large wavelength interval followed by a short continual current dependence of the wavelength. The ECDL in this experiment uses the Littrow configuration, as shown in Fig. 1. This setup includes a collimating lens and a diffraction grating. The firstorder diffracted beam provides optical feedback to the laser diode, which has an anti-reflection coating on the right-hand side. Rotating the grating allows tuning of the emission wavelength. To characterize the light emitted from the external cavity diode laser, we perform several experiments on the laser output. We begin by determining the threshold current, which is detailed in Section II. The second experiment aims to make a precise measurement of the wavelength of the laser (Section III). In the third experiment, we use a monochromator to characterize the power of the laser as a function of wavelength and measure the spectral density (Section IV). In the final experiment in Section V, the wavelength is scanned over a range using a piezo-electric transducer, and the interference fringes on the detector output are observed.

II.

MEASURING THRESHOLD CURRENT

In the first experiment we perform a threshold current measurement of the laser diode. The laser system was started (a more detailed explanation of this procedure can be found in the Appendix), and a power meter was fixed along the beam path (set to its most sensitive setting) to measure the output power of the diode laser. The current, Iset was set to 5 mA, and was gradually increased in steps of 1 mA, until 177 mA. Each time, the output power was recorded. The collected data was plotted and is shown in Figure 2. The threshold for las-

2

FIG. 1: The Littrow configuration

FIG. 3: Setup of the wavemeter adapted from the Burleigh manual.

III.

FIG. 2: Power (mW) vs. Current (mA) graph showing threshold current for lasing.

ing was determined using a two-segment linear fit, where the threshold was calculated to be the point at which a straight-line fit to the linear portion of the P/I curve above the threshold intercepts the straight-line fit to the linear portion of the P/I curve below the threshold. The fit data is shown in Table I. Using the fit data, the intercept was calculated and the threshold current for lasing was determined to be 35.005 ± 0.00074899 mA. Linear Fit Value Standard Error 1 Intercept -0.00418 0.00238 Slope 0.0018 1.11399E-4 2 Intercept -26.90549 0.07426 Slope 0.77029 7.40664E-4 TABLE I: Two-segment fit data for threshold current

MEASURING WAVELENGTH

Characterizing the wavelength of the laser is a critical step in many applications, and we use an interferometric wavemeter to make a precise measurement of the wavelength. The Bristol Instruments 521 Wavemeter was used in this experiment. The setup is shown in Figure 3. Photons enter the wavemeter through a fiber-optic cable and the collimated beam can be coupled into the fiber using a beam receiver (this consists of an input beam coupler and an FC connector-terminated patch cord), which is mounted on a ThorLabs KM-100 optical mount. Since the beam receiver can be adjusted by small angular adjustments in θ and φ, we can optimize the efficiency of the laser light into the fiber. The fiber injects light into the chassis of the wavemeter, inside which is a compact scanning Michelson interferometer. The interferometer splits the incoming laser beam into two paths and varies one by a scan mechanism. The two paths are then recombined, allowing us to observe interference fringes by connecting a photodiode at the output. By combining this signal with the displacement of the scan mechanism captured by a high-res Moire gauge, the microprocessor calculates the wavelength. By using this wavemeter setup, we can calculate the central wavelength of the diode laser. The laser current Iset was adjusted to 50 mA and gradually increased in steps of 5 mA, until a maximum current setting of 150 mA. Each time the current was changed, the wavelength measured by the Bristol software was recorded. Since the incident optical power on the beam receiver should not exceed 10 milliwatts, a power meter was used to measure the laser output onto the receiver in-between current adjustments, and ND filters were used to reduce the power going into the fiber coupler if necessary. The recorded data is shown in Figure 4. The data

3

FIG. 4: Wavelength (nm) vs. Current (mA) graph with a linear fit showing central wavelength of diode laser.

FIG. 5: Block diagram of electronics setup for reading the intensity from the monochromator.

Value Standard Error Intercept 780.21885 0.02489 Slope 8.82353E-5 2.30888E-4

Value Standard Error y0 2.09768 0.01532 0.02766 xc 781.06581 A 3.73498 014898 w 1.49891 0.04897

TABLE II: Linear fit data for central wavelength measured using Bristol Instruments 521 Wavemeter; The first two datapoints have been neglected in the fitting procedure.

was fitted and the the results are shown in Table II.

IV.

MEASURING SPECTRAL DENSITY

In this experiment we use a monochromator to characterize the power of the laser as a function of wavelength. A Newport/ Oriel Cornerstone 260 1/4m system - model 74100 is used, which is connected to a PC using an IEEE488 (or GPIB) cable. Light is collimated and directed into a diffraction grating which is attached to a rotatable mount and a precise gear system. The diffracted light will be detected by a photomultiplier tube (PMT), which amplifies the photon signal as a result of cascading photoelectrons. This signal is then sent to a pre-amplifier which converts it into a voltage, which is proportional to the light intensity for a given wavelength. The complete setup is shown in Figure 5. We used a beam splitter to direct a small fraction of the laser beam through an optical chopper toward the monochromator slits. The optical chopper was set to 850 Hz, and the oscilloscope was set to trigger on the chopper signal. The gate on the SRS Boxcar integrator was adjusted until it overlapped the oscilloscope signal, and further calibrations were done to make sure that closing the shutter on the monochromator would bring the signal on the digital meter to zero.

TABLE III: Gaussian fit data for spectral density distribution

Measurements were taken for the light intensity as a function of wavelength by setting the wavelength interval from 776 to 790 nm in steps of 0.1 nm, which was automated using a modified 74100 LabView control code. At higher light intensities, the signal becomes too large to average properly; therefore, we inserted a neutral density (ND) filter into the beam path before the monochromator. When a saturation region was identified, we added a filter to reduce the power, and repeated data-taking for the region, with several points on either side of the peak to allow for normalization when combining the data. A normalization constant was calculated for each trial using several wavelengths as reference points. The collected data is shown in Figure 6. A Gaussian fit was used, with the equation: y = y0 +

−2(x−xc )2 A p e w2 w π/2

(1)

. We calculated the central wavelength to be 781.068 ± 0.02766 nm, and the spectral bandwidth of the laser was calculated to be 1.49891 ± 0.04897 nm. This measurement is related to the linewidth and is given by the Schalow-Townes equation:

∆Vlaser =

πhV (∆Vcavity )2 Pout

(2)

4

FIG. 6: Voltage vs. Wavelength graph showing spectral density of diode laser.

where ∆Vlaser is the laser bandwidth, hV is the photon energy, ∆Vcavity is the bandwidth of the cavity, and Pout is the output power of the laser during measurement of the cavity bandwidth.

V.

FIG. 7: Oscilloscope trace of interference fringes overlaying the ”sawtooth” scanning wave.

where L1 and L2 are the lengths of the arms of the interferometer. We know that c = f λ, so:

SCANNING THE WAVELENGTH

f=

In the final experiment we scan the wavelength of the laser over a specific range, which is a widely used procedure in most spectroscopic applications. To achieve this, we use the piezo-electric transducer to modify the length of the cavity. We construct a Michelson interferometer using a beam splitter and mirrors, and turn on the scan control (SC 110), and after careful adjustment of the amplitude and scan speed, we can observe the interference fringes by placing a photodiode at the interferometer exit. The alternating bright and dark fringes are a result of constructive and destructive interference of the incoming laser beam and the reflected beam. By measuring the number of interference fringes, we can acquire a tunable wavelength limit using the SC 110. The Michelson interferometer allows the precise measurement of the wavelength of the laser. By scanning the wavelength over a range and counting the interference fringes, we can calculate the wavelength of the diode laser:

λ=

2∆d N

(3)

where λ = wavelength, ∆d = scan distance that causes an N number of fringes to pass through the photodiode.

λ=

2(L1 − L2 ) N

(4)

mc 2(L1 − L2 )

(5)

Therefore:

∆f =

∆mc 2(L1 − L2 )

(6)

The interference pattern observed on the oscilloscope is shown in Figure 7. We measured 24 interference fringes in one scan of the SC 110, with L1 = 100 ± 0.5cm and L2 = 12 ± 0.5cm. Thus, the frequency range, ∆f was measured to be about 4.14 ± 0.5GHz.

VI.

DISCUSSION

The four experiments detailed above were aimed to characterize the light emitted from the diode laser. The diode laser beam diverges as a result of diffraction at the aperture, which in this case is the optical cavity. Typically an edge-emitting diode laser has a beam divergence of 30deg, which indicates that a very small aperture is used. In the first experiment, we measured the threshold current for lasing. The uncertainties in this experiment are largely dominated by the hardware and detector resolutions. At higher currents, we noticed that it would be harder to maintain a constant value. A helpful solution would be to record data twice for current settings greater than 100 mA, which could minimize error due to instability. Also, light is emitted below the threshold frequency

5 as a result of spontaneous emission, and is a source of inefficiency of laser. In the next experiment we determined the central wavelength of the diode laser. Again, the uncertainties are dominated by the accuracy of the wavemeter and the high-res Moire gauge, and the speed with which the microprocessor can calculate the wavelength. It would take a few seconds for the wavelength to stabilize, so care was taken to make sure no data was recorded while the wavelength was still settling. In the third experiment we measured the spectral density of the laser. We calibrated the SRS 250 Boxcar integrator using the ’offset’ setting to make sure that closing the shutter would zero out the signal. The boxcar integrator averages like a charging capacitor; if it is set to 300 triggers (i.e. one RC time constant; this trigger rate is determined by the repetition rate on the chopper), it would take 300 triggers to reach approximately 63% of the maximum value. Therefore in order to get accurate readings, we had to allow for a minimum of a few time constants. During data-taking with filters, we did not foresee the need to measure the same region twice, and this introduced a significant systematic error, because we were forced to normalize the voltage data based on very few points. In the future, the same region must be scanned twice - once with the filter and once without. Only then can the consequent set of voltage values be normalized to construct a spectral distribution. We also noticed that the central

wavelength evaluated using the monochromator was different in comparison to the wavelength measured by the scanning Michelson interferometer in Section III. This is because the interferometer uses a very precise mechanism involving a laser with a specific wavelength in order to make distance measurements. The monochromator, on the other hand, changes the wavelength by varying the angle on a diffraction grating, which is very difficult to calibrate properly. As a result, the central wavelength evaluated using the scanning Michelson interferometer is seen to yield a more accurate measurement. The final experiment was aimed at scanning the wavelength of the diode laser. In order to minimize the uncertainties in the measurements of the interference fringes caused by the physical setup of the Michelson interferometer, we increased L1 such that L1 >> L2 . Unfortunately, the setup was altered before length measurements could be taken; therefore, the values of the interferometer arms were only approximations. Also, an adjustable aperture was used in front of the detector to maximize the contrast and reduce the uncertainty associated with making the oscilloscope measurements. Since we only used an interference pattern during a single scan, future analysis could average over several pulses of the SC 110, which would help reduce the statistical uncertainty on the measured frequency range.

[1] R. Paschotta, article on ’mode hopping’ in the Encyclopedia of Laser Physics and Technology, 1. edition October 2008, Wiley-VCH, ISBN 978-3-527-40828-3. [2] WA-2500 Wavemeter Operating Manual, Burleigh Instruments Inc. (04659-M-00 Rev E), Burleigh Park, Fishers, NY 14453. [3] DL 100 Grating Stabilized Diode Laser Head Manual,

Toptica Photonics (2006). [4] K.B. MacAdam, A. Steinbach, and C. Wienman, Am. J. Phys. 60, 1098 (1992). [5] N. Grau, D. Alton, and E. Wells, Advanced Lab Manual, Augustana College, Sioux Falls (2015).