JP2798350B2 - A method for selecting a waveguide-type electro-optical element in which a drift of a DC voltage is controlled. - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a method for selecting a waveguide type electro-optical element in which a drift of an applied DC voltage is controlled. More specifically, the present invention relates to an electro-optical device, such as a waveguide modulator device made of Ti: LiNbO ₃ , in which the drift of a DC control voltage for controlling the output light is a predetermined value. And a simple method for selecting elements controlled by the method.

[0002]

2. Description of the Related Art As a means for evaluating a drift phenomenon of a control voltage for controlling output light from an electro-optical element, as represented by a waveguide type modulator element made of Ti: LiNbO ₃ , a constant value is used. A method has been reported in which a DC bias voltage is applied to a device to monitor the change over time in the operating point of output light (for example, H. Nagata and K. Kiuch, J. Appl.
hys. 73 (1993) 4162. and K. Seino, T. Nakazawa, Y. Ku
bota, M. Doi, T. Yamaneand H. Hakogi, Proceedings of
the OFC'92, San Jose, February 8-11, 1992 (Optical
Soc. Of America) p.325).

However, in actual use of the device, the operating point of output light is adjusted to a certain initial value by a DC bias voltage, and an AC (high frequency) voltage is superimposed on the operating point to correspond to the external voltage signal. The modulation state of the output light intensity is obtained. For this reason, the DC bias voltage is not fixed to a constant value, but is adjusted and changed every moment so that the operating point of the output light is always maintained at the initial value. This time-dependent change in the bias voltage (control voltage) required to always maintain the operating point of the output light at the initial value is a true DC drift phenomenon. If the change in the operation bias voltage (control voltage) exceeds the allowable value of the system, the element is in an uncontrollable state.

Therefore, in the state where only the fixed DC bias voltage is applied as described above, the D
Even if the C drift is evaluated, it does not correspond to the element operation in the actual system. However, the advantage of the fixed DC bias voltage application method is that regardless of individual differences in the element performance to be evaluated, continuous application of a bias voltage of a specified constant value allows the output light to be obtained even if there are a plurality of samples. By simply monitoring the phase change or the intensity change over time, the change over time of the operating point
It is very easy to measure and evaluate multiple samples with one measurement system.

On the other hand, it is also possible to directly measure and evaluate the drift of the bias voltage (control voltage) generated during the actual operation of the device by using an actual system circuit or a similar circuit (for example, H. Jumonji and T.) .Nozawa, IEICE
Transaction, J75-C-1 (1992) 17). However, the measurement system becomes more complicated than the former, and when evaluating a plurality of samples each having a different drift characteristic, it is necessary to make a one-to-one correspondence between the sample and the control circuit, which further complicates the system.

[0006]

SUMMARY OF THE INVENTION The degree of drift of the control voltage of the element is easily measured, and has a defect characteristic such that a large drift, particularly a control voltage is diverged, as an inspection item during a manufacturing process. Provided is a method for selecting and sifting elements (products). For the above purpose, it is preferable to adopt a conventional fixed DC bias voltage applying method as a basic measuring means in terms of simplicity of measurement and ability of simultaneous evaluation of a plurality. In addition, by clarifying the correspondence between the drift characteristic value measured by the fixed DC bias voltage application method and the drift phenomenon of the control voltage in the actual system, the optical element in consideration of the DC drift generated during the actual operation is considered. Judgment criteria need to be given.

[0007]

Since the DC drift phenomenon is accelerated by increasing the ambient temperature (see the above-mentioned document), the measurement time is reduced by keeping the ambient temperature as high as possible. can do.

Therefore, when operating the waveguide type electro-optic modulator element in a state where a DC bias voltage for adjusting the operating point of the output modulated light to a specified position is applied,
Estimate the degree of `` control voltage drift '' defined as a phenomenon in which the bias voltage required to maintain the operating point at the specified position changes over time, and the control voltage drift does not diverge, that is, it becomes impossible to control. A method for selecting elements that do not need to be applied, wherein a DC bias voltage of an arbitrary constant value is continuously applied to the element maintained at an arbitrary constant temperature environment higher than room temperature (about 30 ° C.) and 100 ° C. or lower. The device is operated at and the change over time of the operating point of the output light (for example, the peak position of the intensity of the modulated light) is also monitored.
The apparent saturation drift amount (VC) obtained in units of voltage from the temporal change of the operating point of the output light, so-called DC drift, and the bias voltage (V
By selecting the elements having a ratio of B) (A = VC / VB) smaller than at least 1, the products can be easily sieved.

By making the heating temperature for accelerating the drift phenomenon higher than room temperature, the measurement time can be shortened. However, in order to suppress the thermal deterioration of the adhesive and the like used for assembling the elements. Preferably, the heating temperature is 100 ° C. or less. When measured at an ambient temperature of 80 ° C., the measurement time required to derive the drift characteristic value “A” was several tens of hours (about one day) in the examples described later.

In the waveguide type electro-optical element according to the present invention, in which the drift of the applied DC voltage is controlled, a desired value of a DC bias voltage is continuously applied to the waveguide type electro-optical element which is maintained at a predetermined temperature equal to or higher than room temperature. In this state, when light is input to the element and the change with time of the operating point of the light output from the element is monitored, the ratio of the saturation voltage drift amount VC to the applied DC bias voltage (VB), A = VC / VB is less than 1.

According to the method of the present invention for selecting a waveguide type electro-optical element in which the drift of the applied DC voltage is controlled, the waveguide type electro-optical element is maintained at a predetermined temperature higher than room temperature and equal to or lower than 100 ° C. A DC bias voltage of a predetermined value is applied to the element, and light is input to the element to which the voltage is applied, the change over time of the operating point of the output light of the element is monitored, and the saturation value (VC) of the voltage drift amount is monitored. And the ratio of the applied DC bias voltage (VB): A = VC / VB,
An element showing a value of less than 1 as the ratio A is selected.

In the above-mentioned element and the method for selecting the same, the value of the ratio A is preferably 0.5 or less, whereby the drift amount of the DC control voltage with respect to the element can be kept low.

[0013]

FIG. 1 shows the DC drift, operating time and (at 70 ° C.) of a waveguide type Mach-Zehnder intensity modulator device made of Ti: LiNbO ₃ measured by a fixed DC bias voltage continuous application method. The applied fixed DC bias voltage was changed to six levels of -5, 0, 1, 2, 4, and 5 V, and the measurement was performed at an ambient temperature of 70 ° C.

The structure of the element used is z-cut LiN
An optical waveguide pattern is formed on the surface of the bO ₃ substrate by thermally diffusing Ti (980 ° C.), and then using a sputtering method to form an SiO ₂ buffer layer (a high-band modulator typically has a thickness of 500 nm to 500 nm). 1 μm) at 600 ° C.
After heat treatment in an oxygen stream, a traveling wave type electrode pattern (A
u) is provided by vapor deposition and electrolytic plating. One of the two ends of the waveguide through a thin-film polarizer,
The optical fiber is fixed with an ultraviolet curing type low refractive index epoxy adhesive. A commercial electrode connector for high frequency is bonded to the input / output section of the electrode pattern.
These components are usually packaged in a metal housing, but in a series of experiments in the present invention, a sample mounted on a housing that is not particularly hermetically sealed was used.

In the measurement of the DC voltage drift in the experiment shown in FIG.
A ± 20 V alternating current voltage of kHz is applied, and the intensity of the intensity-modulated light (wavelength of the input light: 1.55 μm) output from the element is monitored on an oscilloscope, and a specific peak position (for convenience, the applied voltage on the horizontal axis) The change over time of the peak closest to the position of the value OV was tracked (the unit of change is the voltage value (V)). In FIG. 1, the amount of change (unit of voltage) from the initial state is shown as a DC voltage drift amount. A similar drift measurement, without applying an AC voltage,
It can be measured even when only a fixed DC bias voltage is applied. In that case, the drift is measured as the output light intensity change, but if the initial operating point of the element and the half-wavelength voltage are known, it is easy to convert the measured intensity change into a drift amount in voltage units. .

FIG. 2 shows that at a temperature of 70 ° C., first, a fixed DC bias voltage (−5, 1, 4 V: indicated by white symbols)
DC voltage drift is measured with
Next, at the stage when the drift value is almost saturated, the drift value (black symbol) measured by setting the applied bias voltage to zero is shown. From FIG. 2, it can be clearly understood that when the applied DC bias voltage is turned off, the DC drift value caused by the previous bias application recovers in substantially the same time.

The results shown in FIGS. 1 and 2 are similar to the polarization phenomenon when an electric field (bias) is applied to the dielectric, particularly the orientation polarization having a relatively long relaxation time or the ionic polarization phenomenon. . Although the observed DC drift is not always due to the contribution of polarization, the materials constituting the device are oxide dielectrics and ferroelectrics, and the observed drifts may be as long as several hours. Considering that the phenomenon is a necessary phenomenon, it can be inferred that the phenomenon is at least due to ion displacement / migration, not a phenomenon due to the movement of electrons in the device.

Furthermore, the drift depends on the direction of the bias, and the drift recovers when the bias voltage is cut. Therefore, the displacement of ions or ion pairs, that is, the polarization phenomenon, is more involved than the ionic conduction mechanism. it is conceivable that. That is, by applying the bias voltage, polarization occurs in a direction to cancel the electric field (bias) with a time delay (relaxation time), and the bias voltage effectively applied to the element gradually decreases. This seems to be observed as a DC voltage drift phenomenon. Although the polarization mechanism is not clear at this stage,
It is considered that ions constituting the dielectric itself are generated in —OH groups based on hydrogen contained as impurities in the oxide material, and water molecules contained as impurities in the element.

Therefore, the voltage drift phenomenon having a time delay observed when a single constant value DC bias voltage is continuously applied as shown in FIGS.
It can be expressed by a general formula relating to the relaxation phenomenon, V (t) = V (∞) ｛1−exp (−t / τ)｝ (0). In the above equation, t represents time, and τ represents relaxation time.

In order to confirm this, FIG.
Of the operation time and log (V (∞) −V (t))
And plotted. V (∞) is the apparent saturation drift voltage, and V (t) is the drift voltage at time t. From FIG. 3, the DC voltage drift phenomenon observed in the fixed DC bias voltage method can be expressed by equation (0), and the relaxation time (at an ambient temperature of 70 ° C.) is shown in parentheses in FIG. As noted, it was found to be about 4 hours.

Next, D measured by the fixed bias voltage method
Based on the above consideration on the C drift phenomenon, the measurement data by the fixed bias voltage method was associated with the drift of the control voltage in an actual system. In actual device operation, first, an initial bias voltage is applied to adjust the operating point of the device to an optimal voltage position. When a drift V (Δt) occurs after a time Δt due to this initial bias voltage, the voltage V (Δt) corresponding to the drift is added to the initial bias voltage. By repeating such an operation, the operating point (modulation state) of the output light can always be kept the same as the initial state.

In the following, based on the drift characteristic value obtained by the fixed bias voltage method, actual device operation (following
It is possible to derive a relational expression for estimating the drift of the control voltage appearing in the wing up control method.

At time t = 0, a DC bias voltage V
_{When B} is applied, the DC drift V at time t
(T) is given by the following equation (1) if the ambient temperature is constant:

[0024]

(Equation 1) [Where A: coefficient (V (∞) = AV _B ) τ: relaxation time].

When the applied bias voltage V _B changes with time, it is assumed that the total drift V (t) is given by the sum of the drifts for each additional bias based on the equation (1). From the above assumption, when the initial bias voltage V ₀ is applied to the element, at time t = nΔt,
A value of a bias voltage (BnΔt) to be added to maintain the operating point of the output light from the element in the same state as the initial operating point is obtained.

As shown in FIG. At time t = 0, an initial bias voltage V ₀ is applied. 1. Drift V (Δt) at time t = Δt is obtained by the following equation.

[0027]

(Equation 2) 1 '. At time t = Δt, bias voltage B (Δt) = V (Δ
t) is additionally applied. 2. Voltage drift value at time t = 2Δt: V (2Δt)
Is V ₁ + V ₂ as shown in FIG. That is, it is obtained by the following equation.

[0028]

(Equation 3) 2 '. The bias voltage B (2
Δt) is obtained by the following equation, as shown in FIG.

[0029]

(Equation 4) Similarly, bias voltages B (3Δt) and B (4Δt) at times t = 3Δt and t = 4Δt are obtained by the following equations.

[0030]

(Equation 5)

Therefore, the bias voltage B (nΔt) at the time t = nΔt is obtained by the following equation (2).

[0032]

(Equation 6)

The drift amount V (nΔt) of the control voltage at time t = nΔt is obtained by the following equation (3) based on the above equation (2).

[0034]

(Equation 7)

In order for V (nΔt) to converge, it is necessary that the following relational expression (4) is satisfied: B (nΔt) <B ((n−1) Δt) (4) Therefore, when conditions for satisfying this relationship are obtained, B (nΔt) and B ((n−
1) The difference from Δt) is obtained from the following equation (5).

[0036]

(Equation 8)

From equation (5), it can be seen that if A <1, the relationship of equation (4) is satisfied and V (nΔt) converges. Also, when A = 1, B (nΔt) = B ((n−1) Δt) according to equation (5), that is, V (nΔt) diverges linearly. From equation (3), the convergence state of V (nΔt) is
It can be seen that it depends on the parameters A, Δt, and τ.

From the equation (3), the drift amount of the control voltage at an arbitrary time can be calculated by the control interval Δt, the relaxation time τ measured by the fixed bias voltage method, and the apparent saturation drift amount (voltage) measured by the fixed bias voltage method. (Unit) applied to the applied bias voltage. The condition for the control voltage drift not to diverge but to converge is that the ratio A <1 according to equation (5), that is, the apparent saturation drift amount (VC) obtained by the fixed bias method is equal to the applied bias voltage (VC). VB). Δt
Is a system dependent parameter. τ is a temperature-dependent parameter.If the temperature acceleration coefficient is known in advance, the drift amount at an arbitrary temperature can be estimated by multiplying the value of τ measured in a high-temperature acceleration environment by the acceleration coefficient. Can also. At least A <1 was shown as a reference value for selecting the drift characteristics of the elements. However, it is clear that the element having a larger τ has a slower drift than the element. Therefore, it is of course possible to use the value of τ as a criterion for element selection.

FIG. 5 shows the result of calculation by substituting actual values into the equation (3). According to the measurement examples of FIGS. 1 and 3, the value of τ was set to 4 hours. The initial bias voltage is 3.5 V corresponding to the half-wavelength voltage of the element, the value of Δt is 1 hour for convenience, and the value of A is 1, 0.8, 0.5, 0.2, 0.1. Calculated for the case. In the case of A = 1, the required control voltage (VB) linearly increases and diverges as the operation time elapses, as can be seen from Expression (5).

FIG. 6 shows the relationship of FIG.
It has been rescaled. As can be seen from FIGS. 5 and 6, the control voltage does not diverge if A <1, but in order to operate the device within a lower control voltage range, particularly within 10 V (including the initial bias), A ≦ 0.
It is preferably 5.

FIG. 7 shows that the value of Δt was 5 to 5 under the conditions of an initial bias voltage: 3.5 V, a relaxation time: 4 hours, and A = 0.5.
The relationship between the operation time and the operation voltage when the calculation is performed by changing the change in the range of 240 minutes is shown. As Δt increases, at least the initial drift voltage decreases. Therefore,
When estimating the absolute amount of the control voltage drift from Expression (3), it is safe to make Δt as small as possible. Assuming that Δt → 0, Expression (3) may be integrated to derive a general expression or an approximate expression.

[0042]

EXAMPLE The correspondence between the actual measurement example and the equation (3) was examined. In the device sample shown in FIG. 1, the value of A is almost 1, and
In order to exhibit a divergent type drift, an element having A <1 was manufactured. This device had the same form as that shown in FIG. 1, but the SiO ₂ buffer layer was produced not by sputtering but by vacuum evaporation.

FIG. 8 shows a DC drift measured by a fixed bias voltage method at an ambient temperature of 80 ° C. and a bias voltage of 5 V. The drift initially occurred in the negative direction, and turned into a drift in the positive direction from the middle as shown in FIG. Although the cause of the initial generation of the drift in the negative direction is unknown, it was considered that the drift presumably due to the polarization phenomenon or the like would eventually become dominant, and the correspondence with the equation (3) was examined. Since the drift occurs once in the negative direction, as an apparent saturation drift amount, as shown in FIG.
0.28) and 2.3 V (A = 0.46).

The relaxation time of the drift in FIG. 8 was 8.4 hours as shown in FIG. (Reciprocal of the plot slope).

FIG. 10 is a diagram similar to the actual system operation.
That is, it shows the correspondence between the drift value of the control voltage measured by a method of controlling the bias voltage as needed (at 5-minute intervals) to compensate for the change of the operating point due to the drift, and the result calculated by the equation (3). The straight line is a result obtained by calculating Δt = 1 hour under the condition of A = 0.28 when the relaxation time = 8.4 hours and A = 0.46. The initial bias voltage was 3.5 V in both the actual measurement and the calculation. Considering that Equation (3) is a first approximation derived based on many assumptions, the calculation and the measured value in the case of A = 0.28 show relatively good agreement.

In the case where there are two ways of determining the value of A as in this example, the threshold for element selection is determined under the condition that the control voltage upper limit is determined. It is preferable to make some preliminary measurements.

[0047]

According to the present invention, there is provided a waveguide type electro-optic modulator element having a preferable DC voltage drift characteristic and a method of selecting the same.

[Brief description of the drawings]

FIG. 1 is a graph showing an example of DC drift of a waveguide type Mach-Zehnder intensity modulator device made of Ti: LiNbO ₃ measured by a fixed bias continuous application method.

FIG. 2 shows an example of a DC voltage drift in a state where a fixed DC bias (−5, 1, 4V: indicated by an open symbol) is continuously applied at a temperature of 70 ° C. The graph which shows an example of the drift (black symbol) measured by making the applied bias zero at the stage of saturation.

FIG. 3 is a diagram showing the data of FIG.
A graph plotted by changing the relationship of (V (∞) -V (t)).

FIG. 4 is a graph showing the relationship between a drift phenomenon of a control voltage and a drift phenomenon observed in a fixed bias voltage method, and an actual element control method.

FIG. 5 is a graph showing a relationship between an operation time and an operation voltage according to an equation (3).
Is a graph showing the result of calculation by actually substituting values for.

FIG. 6 is a graph in which the operation time is changed to a log scale in the relation between the operation time and the operation voltage in FIG. 5;

FIG. 7 is a relationship between operation time and operation voltage when the value of Δt is changed from 5 to 240 minutes under the conditions of an initial bias of 3.5 V, a relaxation time of 4 hours, and A = 0.5. A graph showing.

FIG. 8 shows an atmosphere temperature of 80 ° C. and a bias voltage of 5 V.
5 is a graph showing an operation time-DC drift relationship when measured by the fixed bias method of FIG.

FIG. 9 is an operation time for calculating a relaxation time of DC drift shown in FIG. 8, which is -ln ｛V (∞) -V;
(0) Graph showing the relationship.

FIG. 10 shows a drift of the control voltage measured in a manner close to the actual system operation, that is, a method of controlling the bias voltage as needed (at 5-minute intervals) to compensate for a phase change due to the drift, and an equation ( 3 is a graph showing an operating time-operating voltage relationship for indicating the correspondence between the results calculated in 3).

──────────────────────────────────────────────────続 き Continuing from the front page (72) Inventor Junichi Ogihara 585 Tomicho, Funabashi-shi, Chiba Sumitomo Cement Co., Ltd. Central Research Laboratory (72) Inventor Junichiro Minowa 585 Tomicho-cho, Funabashi-shi, Chiba Sumitomo Cement (56) References JP-A-5-257105 (JP, A) JP-A-4-346310 (JP, A) JP-A-5-215927 (JP, A) OPTICAL FIBER COM MUNICATION CONFERENCE, 1992 TECHNICAL DIGEST SERIES, VOL. 5, PP. 325-328 (FEBRUARY 2-7, 1992). SEINO ET. AL. , "A LOW DC-D RIFT TI: LINBO3 MOD ULATOR ASSURED OVER R 15 YEARS" (58) Fields investigated (Int. Cl. ⁶ , DB name) G02F 1/00-1/055 505 G02F 1/29-1 / 313 G02B 6/12-6/14

Claims (2)

(57) [Claims] 1. A plurality of waveguide-type electro-optical elements are maintained at a predetermined temperature higher than room temperature and equal to or lower than 100 ° C., and a DC bias voltage having a predetermined value is applied to these elements. Light is input, the change over time in the operating point of the output light of this element is monitored, and the saturation value (V
C) and calculating the ratio of the applied DC bias voltage (VB): A = VC / VB, and selecting an element having a value of less than 1 as the ratio A. For selecting a waveguide-type electro-optical element in which drift of the waveguide is controlled. 2. The method according to claim 1, wherein the value of the ratio A is 0.5 or less.