Covesion specialises in the manufacture of periodically poled lithium niobate (PPLN) devices, such as, MgO-doped periodically poled lithium niobate (MgO:PPLN or PPMgO:LN) and undoped PPLN. These PPLN devices are used for efficient frequency conversion of lasers allowing you to reach wavelengths that cannot be achieved with conventional solid state lasers, diode lasers etc. For example, you can use PPLN to:

• frequency double a 1064nm laser to 532nm – a technique used for green laser pointers
• convert 1064nm to 3um, used for gas detection or microscopy imaging techniques

Principles of nonlinear frequency conversion

• Second order nonlinear processes
• Phase matching of MgO:PPLN with periodic poling period

Read More

Material properties of Lithium Niobate

• Second Order Nonlinearity
• Refractive index
• Transmission
• MgO:PPLN vs undoped PPLN
• Power Handling and Damage Threshold
• Damage Mechanisms
• The Photorefractive Effect
• Green Induced Infrared Absorption

Read More

How to use PPLN

• Crystal length
• Polarization
• Focusing and the Optical Arrangement
• Temperature and Period

Read More

Example uses of PPLN

• Second Harmonic Generation
• Difference Frequency Generation
• Optical Parametric Oscillator
• Sum Frequency Generation

Read More

Second Harmonic Generation

PPLN can be used in a single pass configuration for SHG with the pump focussed at the centre of the crystal length. For optimum efficiency, aim for the Boyd-Kleinman focussing condition. This is where the spot size is such that the ratio of the crystal length to the confocal parameter is 2.84.

The optimum conversion efficiency that can be achieved for an SHG interaction also depends on several factors such as:

• CW or pulsed pump source
• Input power: at high power, you can reach gain saturation
• Pump/SHG wavelength: At low gain, conversion efficiency is higher for interactions involving higher energy photons (short wavelength).

1064nm → 532nm

For low gain CW the typical conversion efficiency is 2%/Wcm. For example, for 1.5W at 1064nm and a 40mm long MgO:PPLN crystal, the expected 532nm output is 180mW. At higher powers, Covesion has achieved 1.5%/Wcm with a 10W source, generating 3W at 532nm from a 20mm long crystal.

In CW systems, conversion efficiencies in excess of 50% have been demonstrated in an intracavity arrangement^[1]. For nanosecond sources (~10KHz, ~50uJ), efficiencies of 50% can typically be achieved.

1550nm → 775nm

Frequency doubling of Erbium doped fiber lasers is also common, for example for 775nm or 780nm generation. For a CW source, you can typically achieve 0.6%/Wcm for low gain. At high powers an efficiency of 0.3%/Wcm has been demonstrated for generating 11W at 780nm in a 40mm long crystal with 30W pump power^[2].

For a nanosecond source, up to 80% conversion efficiency has been demonstrated in a single pass pulsed system^[3]. For femtosecond sources, using a 1mm crystal length, customers have reported efficiencies of 40-60% for ~100fs, 100MHz and several hundred mW average powers. Due to the very wide temperature acceptance bandwidth, our MSHG1550-0.5-1 crystal can be used at room temperature, and with no temperature controller, for SHG at 1550 or 1560nm.

Difference Frequency Generation

PPLN is often used in a DFG setup for mid-IR generation, either with a tuneable Ti:S laser and 1550nm laser, or a 1064nm source and tuneable ~1550nm laser. Optimum efficiency requires confocal focussing of both pump beams, i.e. ratio of the crystal length to the confocal parameter is 1. For CW systems, efficiencies of 0.3-0.4mW/W2cm can be achieved.

Optical Parametric Oscillator

One of the most common uses of PPLN is in an Optical Parametric Oscillator (OPO). A schematic of an OPO is shown above. The common arrangement uses a 1064nm pump laser and can produce signal and idler beams at any wavelength longer than the pump laser wavelength. The exact wavelengths are determined by two factors: energy conservation and phase matching. Energy conservation dictates that the sum of the energy of a signal photon and an idler photon must equal the energy of a pump photon. Therefore an infinite number of generated photon combinations are possible. However, the combination that will be efficiently produced is the one for which the periodicity of the poling in the lithium niobate creates a quasi-phase matched condition. The combination of wavelengths that is quasi-phase matched, and hence referred to as the operation wavelength, is altered by changing the PPLN temperature or by using PPLN with a different poling period. Nd:YAG pumped OPOs based on PPLN can efficiently produce tunable light at wavelengths between 1.3 and 5μm and can even produce light at longer wavelengths but with lower efficiency. The PPLN OPO can produce output powers of several watts and can be pumped with pulsed or CW pump lasers.

The minimum oscillation threshold can be achieved under confocal focussing conditions for the pump and resonating signal or idler, i.e. ratio of the crystal length to the confocal parameter is 1. The typical pump threshold for a singly resonant CW OPO is around 1-2W.

Sum Frequency Generation

To achieve efficient SFG, you ideally want the two pump beams to be confocally focussed into the PPLN (i.e. ratio of the crystal length to the confocal parameter is 1) and for both beams to be roughly equal in power.

SFG in PPLN is often used for laser cooling of atoms or ions where very precise control over the frequencies is required. For generation of 626nm light from 1051nm and 1551nm, efficiencies of 3.5-2.5%/Wcm have been achieved. Here, the efficiency η, is defined by^{[4, 5]}:

Where P is the power at each wavelength, and l is the crystal length. An efficiency of 44% has been demonstrated for the generation of 7.2W of 626nm light from 1051nm (8.5W) and 1551nm (8.3W) [4].

A similar conversion efficiency of 3.2%/Wcm has also been reported for 589nm generation from 1064nm and 1319nm^[6].

References

1. M.Zhou et al., Laser Physics, vol. 20, no. 7, pp. 1568-1571 (2010)
2. S. S. Sané et al., Optics Express, vol. 20, no. 8, pp. 8915–9, (2012)
3. D. Taverner et al., Optics Letters, vol. 23, no. 3 pp. 162-164 (1998)
4. H.-Y. Lo et al., Applied Physics B, doi:10.1007/s00340-013-5605-0, (2013)
5. A. C. Wilson et al., Applied Physics B, vol. 105, no. 4, pp. 741–748, (2011)
6. J. Yue et al., Optics letters, vol. 34, no. 7, pp. 1093–5, (2009)

To get the most out of our PPLN crystals, there are four key aspects that you need to consider:

• Crystal length
• Polarization
• Focusing and the optical arrangement
• Temperature and period

Crystal length

Each crystal has an associated pump acceptance bandwidth which is inversely dependent on length, so crystal length is an important factor when choosing a crystal. This acceptance bandwidth is due to the group velocity mismatch between the interacting waves.

For narrowband CW sources our longer crystal lengths, at 20 to 40mm, should give best efficiency. However, for pulsed sources, a long crystal can have a negative effect if the pump bandwidth is much broader than the crystal acceptance bandwidth. For nanosecond pulses, we typically recommend 10mm lengths and our shortest lengths at 0.5 to 1mm are ideal for femtosecond pulse systems.

For SHG of femtosecond pulses, if the pump bandwidth is significantly wider than the acceptance bandwidth, it is still possible to achieve high conversion efficiency. The pump frequencies outside of the acceptance bandwidth can still contribute to the conversion efficiency via sum frequency generation, essentially squeezing the broadband pump into a relatively narrower-band SHG pulse [1].

Polarization

In order to access the highest nonlinear coefficient of lithium niobate, the input light must e-polarized, i.e. the polarization must be aligned with the dipole moment of the crystal. This is accomplished by aligning the polarization axis of the light parallel to the thickness of the crystal. This applies to all nonlinear interactions.I or type II interactions, for example for entangled photon systems for the generation of orthogonally polarized pairs.

This configuration is known as Type-0 phase matching (ee-e), as all the interacting beams have the same polarization.

Type I phase matching (oo-e) and type II phase matching (eo-e) schemes are also possible in PPLN, for example for the generation of heralded single photons. Please contact Covesion to discuss your requirements.

Focusing and the Optical Arrangement

Typically, Covesion crystals consist of several grating periods each with a 0.5×0.5mm2, or 1.0×1.0mm2 aperture and with a length of up to 40mm long. To achieve high conversion efficiency in PPLN, the pump beam should be focussed into a grating with the focus centred on the crystal length.

For SHG with CW lasers, a theoretical result from Boyd and Kleinmann shows that optimum efficiency can be achieved when the ratio of the crystal length to the confocal parameter is 2.84 [2]. (The confocal parameter is twice the Rayleigh range). This is also true for SFG interactions where the two pump beams should also have the same Rayleigh range.

For DFG and OPOs, optimum efficiency requires a confocal focussing condition i.e. the Rayleigh range is half the length of the crystal.

These focussing conditions apply to pulsed lasers too, but due to the high peak powers, the spot size requirements are less sensitive. (Be aware of the crystal damage threshold so as not to focus too tightly.)

In general, a good rule of thumb is that the spot size should be chosen such that the Rayleigh range is half the length of the crystal. The spot size can then be reduced in small increments until the maximum efficiency is obtained.

Temperature and Period

The poling period of a PPLN crystal is determined by the wavelengths of light being used. The quasi-phase-matched wavelength can be tuned slightly by varying the temperature of the crystal.

Covesion’s range of off-the-shelf PPLN crystals each include multiple different poling periods, which allow different wavelengths to be used at a given crystal temperature. Our calculated tuning curves give a good indication of the required temperature for phase-matching. The temperature dependence of conversion efficiency follows a sinc ² function, describing a crystal temperature acceptance bandwidth. The longer the crystal, the narrower and more sensitive the acceptance bandwidth.

In many cases the efficiency of the nonlinear interaction is very sensitive to <1°C. For example, for SHG with a 1064nm pump in a 20mm long crystal, the temperature acceptance bandwidth is ~1°C. So if the temperature is 0.5°C off from the optimum phase matching temperature, then the SHG power is 50% lower than the optimum. If the crystal temperature can be maintained at the optimum phase matching temperature to within +/-0.1°C, then the SHG power is stable to within 2-3%.

The optimum temperature can be determined by heating the crystal to 20°C higher than the calculated temperature and then allowing the crystal to cool whilst monitoring the output power at the generated wavelength.

The Covesion PPLN oven is easy to incorporate into an optical setup. It can be paired with Covesion’s OC1 temperature controller to maintain the crystal temperature to within ±0.01°C, providing highly stable output power.

References

1. K. Moutzouris et al., Optics letters, vol. 31, no. 8, pp. 1148–50, (2006)
2. G. Boyd and D. Kleinman, Journal of Applied Physics, vol. 39, no. 8, p. 3597, (1968)

Second Order Nonlinearity

The second order nonlinear polarisation of lithium niobate can be written as,

The 2-D matrix describes the non-susceptibility tensor χ(2). For 5% MgO doped lithium niobate (MgO:LN) at 1064nm, d31=4.4pm/V, d33=25pm/V^[1].

The highest nonlinear coefficient is d33=25pm/V, which corresponds to interactions that are parallel to the z-axis, i.e. type-0 phase matching. In other words, all interactive waves must be e-polarized in order to achieve the highest conversion efficiency. All of our crystals are designed to access this d33 coefficient. For periodically poled MgO:LN, the effective nonlinear coefficient deff is typically 14pm/V.

NOTE: Covesion can offer custom crystals for type I or type II interactions, for example for entangled photon systems for the generation of orthogonally polarized pairs.

Refractive Index

The temperature dependent refractive index is described by the Sellmeier equation:

Where the temperature dependent parameter f is defined as,

Where T is temperature in °C

and the Sellmeier coefficients are, an optical resonator, also known as an optical parametric oscillator (OPO), the efficiency can be significantly enhanced.

Using these parameters in the Sellmeier equation, you can calculate the refractive index variation with wavelength and temperature. The table below has a few examples.

PPLN has a high index of refraction that results in a ~14% Fresnel loss per uncoated surface. To increase transmission through our crystals, the crystal input and output facets are AR coated, thus reducing the reflections at each surface to less than 1%.

Transmission

MgO:LN and LN have very similar transmission curves and are highly transparent from 400-4000nm. Material absorption occurs below 400nm and above 4000nm where PPLN can still be used as long as the losses can be overcome. For example, a pulsed mid-infrared OPO generating 7.3um has been demonstrated in PPLN^[4], although more commonly, PPLN-based OPOs are often operated up to 4.5-5um. Similarly, for the UV region, generation at 386nm^[5] and 370nm^[6] has been demonstrated using 3^rd order QPM in MgO:PPLN.

Work by Schwesyg et al. have analysed the absorption losses of MgO:LN between 300 and 2950nm^[7]. Their data (shown below) provides an accurate measurement of the absorption coefficient between 400-800nm. Their experiment also found no measurable absorption bands between 800-2000nm.

The figure below shows the transmission curves of LN and MgO:LN measured by Covesion, showing the roll-off of transmission for both materials. The measurement includes Fresnel reflections off both input and output facets of the measured samples, which accounts for a loss of approximately 30% due to Fresnel reflections.

NOTE: There is an OH-absorption band at 2826nm with a measured absorption coefficient of 0.088cm-1 [7].

MgO:PPLN vs undoped PPLN

Undoped PPLN is usually operated at temperatures between 100°C and 200°C, to minimize the photorefractive effect that can damage the crystal and cause the output beam to become distorted. Since the photorefractive effect is more severe in PPLN when higher energy photons in the visible part of the spectrum are present, it is especially important to use the crystal only in the recommended temperature range.

The addition of 5% MgO to lithium niobate significantly increases the optical and photorefractive resistance of the crystal while preserving its high nonlinear coefficient. With a higher damage threshold, MgO:PPLN is more suitable for high power applications. It can also be operated from room temperature up to 200°C, significantly increasing the wavelength tunability of the device. Moreover, in some special cases, the MgO:PPLN can be operated at room temperature and without the need for temperature control e.g. Our MSHG1550-0.5-1 (1mm long) can be used for generating 780nm from 1560nm femtosecond fiber laser.

Power Handling and Damage Threshold

Lifetime testing of our crystals is an on-going process at Covesion. Using a 10W 1064nm CW laser, we have generated 2.2W at 532nm. With a pump intensity of >500KW/cm2 and operating temperature of 35degC, our PPLN maintained the 2.2W SHG output power over a period of 2000hrs, with no signs of damage to the crystal and no evidence of beam distortion due to photorefraction.

The damage threshold of MgO:PPLN or PPLN depends on wavelength as well as whether the source is CW or pulsed. In the CW regime, the threshold depends on the intensity and is lower when visible wavelengths are involved. For pulsed sources, the damage threshold depends on wavelength, pulse duration, average power and the repetition rate. Often the damage threshold will be higher for low repetition rate sources.

If you think that you are working close to the damage threshold, then a good tip is to test the damage threshold in an unpoled region of the crystal. Covesion crystals have a standard width of 10mm, but the poled gratings cover a maximum width ~7mm. You can use the unpoled areas to carefully test for damage as long as it is still within the AR coated region.

Note: The damage threshold in a poled region will be lower if you are generating visible wavelengths. Always increase the pump power gradually, whilst monitoring the beam for any distortions or a sudden drop in power.

The table below shows a collection of data from Covesion and from customers showing the power handling or damage thresholds under various regimes. We are continuously working together with our customers to increase the amount of information available on crystal damage thresholds.

Damage Mechanisms

The Photorefractive Effect

Under conditions of high intensity, LiNbO3 and MgO: LiNbO3 are prone to the photorefractive effect, which is an optically induced change in refractive index. (N.B. The threshold is higher for MgO: LiNbO3).

In a region of high optical intensity, electrons are released as free carriers and then redistribute in an area of lower optical intensity. This causes a spatially varying refractive index within the material that can be observed as beam distortions. This can result in permanent damage to the crystal. However, under some circumstances, if the effects are small the damage can be reversed by heating the crystal to 200°C for a couple of hours to allow all the charge carriers to re-diffuse.

If you are working near the damage threshold, it is recommended that you operate at high temperatures between 150-200°C.

Green Induced Infrared Absorption

Green Induced Infrared Absorption, or GRIIRA, is an effect where the presence of green light allows infrared to be absorbed. This causes local heating which can offset the phase matching temperature of your interaction, but it can also eventually lead to crystal damage.

The mechanism for GRIIRA comes from the creation of polarons from crystal defects such as, Nb ions occupying Li ion sites (known as antisite defects), and Fe ion impurities. Doping lithium niobate with MgO reduces the onset of GRIIRA, as it allows Mg ions to replace the Nb antisite defects.

IR absorption due to blue light also occurs by the same mechanisms, and is known as BLIIRA (blue induced infrared absorption).

References

1. Compact blue-green lasers”, W. P. Risk, T. R. Gosnell and A. V. Nurmikko, Cambridge University Press, 2003
2. Gayer et al., Applied Physics B 91, 343-348 (2008)
3. D.H.Jundt, Optics Letters V.22 N.20 p.1553-1555 (1997)
4. M. A. Watson et al., Optics Letters, vol. 27, no. 23, pp. 2106–8, ( 2002)
5. R.T. White et al., Applied Physics B: Lasers and Optics, 77(6-7), 547–550 (2003)
6. J. Kim, et al., 2013 IEEE Photonics Society Summer Topical Meeting Series (pp. 183–184) (2013)
7. J. R. Schwesyg et al., Advances in Optical Materials, AIThE3, (2011)
8. S. S. Sané et al., Optics Express, vol. 20, no. 8, pp. 8915–9, (2012)
9. F. Kienle et al., Optics Express, vol. 18, no. 8, pp. 7602–10, (2010)
10. F. Kienle et al., Journal of the Optical Society of America B, vol. 29, no. 1, p. 144, (2011)
11. P. K. Upputuri and H. Wang, Applied Physics B, vol. 112, no. 4, pp. 521–527, (2013)

When light travels through a material, it interacts with it on an atomic and molecular level. You can think of these atoms or molecules as arrays of dipoles. The electric field from the incident light drives these dipoles causing them to oscillate like springs as it travels through the material.

In most cases, the light will be unaffected and have exactly the same frequency when it leaves the medium. However, it is possible for the light to force these dipoles to the point that they oscillate with a nonlinear response such that the re-emitted light contains additional frequencies, like the harmonics on a spring. Some materials are more prone to exhibit second order nonlinear or χ ⁽²⁾ responses, others can be more susceptible to third-order or χ ⁽³⁾ responses. The type of nonlinear response depends wholly on the structure of the material.

Second order nonlinear frequency conversion

Second order nonlinear processes involve the mixing of three electromagnetic waves, where the magnitude of the nonlinear response of the crystal is characterized by the χ ⁽²⁾ coefficient. This can give rise to the following interactions:

• Difference frequency generation (DFG)
• Second harmonic generation (SHG)
• Sum frequency generation (SFG)

Second harmonic generation (SHG), or frequency doubling, is the most common application that utilizes the χ ⁽²⁾ properties of a nonlinear crystal. In SHG, two input pump photons with the same wavelength λ _p are combined through a nonlinear process to generate a third photon at λ_SHG, where,

OR, in terms of frequency,

Similar to SHG, sum frequency generation (SFG) combines two input photons at λ _p and λ_s to generate an output photon at λ _SFG , where,

OR, in terms of frequency,

Alternatively, in difference frequency generation (DFG) when two input photons at λp and λs are incident on the crystal, the presence of the lower frequency signal photon, λs, stimulates the pump photon, λp, to emit a signal photon λ _s and idler photon at λ _i , where,

OR, in terms of frequency,

In this process, two signal photons and one idler photon exit the crystal resulting in an amplified signal field. This is known as optical parametric amplification. Furthermore, by placing the nonlinear crystal within an optical resonator, also known as an optical parametric oscillator (OPO), the efficiency can be significantly enhanced.

Phase Matching

In all of these processes, photon energy is conserved; however in order for any of these the second order nonlinear frequency conversion interactions to occur, momentum must also be conserved. This is otherwise known as phase matching.

Phase matching refers to fixing the relative phase between two or more frequencies of light as they propagate through the crystal, such as frequency doubling, sum and difference frequency generation . The refractive index is dependent on the frequency of light. Thus, the phase relation between two photons of different frequencies will vary as the photons propagate through the material, unless the crystal is phase matched for those frequencies. It is necessary for the phase relation between the input and generated photons to be maintained throughout the crystal for efficient nonlinear frequency conversion of input photons. If this is not the case, the generated photons will move in and out put phase with each other in a sinusoidal manner, limiting the number of generated photons that exit the crystal. This is shown in the figure below. Traditional phase matching requires that the light is propagated through the crystal in a direction where the natural birefringence of the crystal matches the refractive index of the generated light. Despite providing perfect phase matching, this technique is limited to a small range of wavelengths in those materials that can be phase matched.

PPLN is an engineered, quasi-phase-matched material. The term engineered refers to the fact that the orientation of the lithium niobate crystal is periodically inverted (poled). By inverting the crystal orientation at every peak of the sinusoidal generation, one can avoid the photons slipping out of phase with each other. As a result, the number of generated photons will grow as the light propagates through the PPLN, yielding a high conversion efficiency of input to generated photons (see above figure).

The period with which the crystal needs to be inverted (the poling period) depends on the interacting wavelengths and the temperature of the PPLN. For example, a PPLN crystal with a poling period of 6.6μm will efficiently generate frequency doubled photons from 1060nm photons when the crystal temperature is held at 100°C. By increasing the temperature of the crystal to 200°C the same PPLN crystal will efficiently generate frequency doubled photons from 1068.6nm wavelength photons. Thus, changing the temperature of the crystal therefore varies the phase matching conditions, allowing some tuning of the wavelength interaction.