RU2253168C1 - Method for sorting out semiconductor devices - Google Patents

Abstract

FIELD: microelectronics; testing and checking semiconductor devices.

SUBSTANCE: noise-factor is measured on sampled semiconductor device at currents up to 1 mA, ampere-noise characteristics are constructed for maximal and minimal values, current causing highest difference in ampere-noise characteristics is evaluated, and mean noise-factor value is found for given currents. is used for sorting out Lot of instruments is sorted out with respect to their reliability by deviation of noise-factor of each device from mean value at given current.

EFFECT: enhanced reliability and enlarged functional capabilities without bringing in uncontrolled faults.

1 cl, 1 dwg

Description

The invention relates to microelectronics, and in particular to the field of testing and control of semiconductor devices (PP (diodes and transistors)), and can be used for their sorting by potential reliability, as well as to increase the reliability of other methods of sorting both during production and incoming control at manufacturers of electronic equipment.

A known method of sorting PP using various external influences (high and low temperature, electrical loads, etc.) based on heating, cooling the test object, passing an electric current, followed by measurement of parameters [1].

The closest analogue is a method for rejecting potentially unreliable integrated circuits (ICs) on bipolar structures using an integrated current-voltage characteristic [2]. The disadvantage of this method is that it allows you to reject IP and bipolar transistors with latent structural defects only due to contamination of the crystal surface of the product.

The invention is aimed at increasing the reliability of the rejection of software and expanding the functionality without introducing uncontrolled defects.

.This is achieved by the fact that ampere-noise characteristics are measured on a representative sample of the PP batch, which must be divided by reliability into two groups, at small current values (for example, up to 1 mA) and the maximum variation in ampere-noise characteristics for the measured devices is determined the value of the current at which the spread of the noise factor will be the largest, and find the average value of the noise factor at this current

.

Further, by measuring the noise factor of each device at the current value U ² _{sh found} , according to the criteria U ² _sh ≥ U ² _sh.cp , and U ² _sh <U ² _sh.cp , the batch is divided into less reliable and reliable devices, respectively.

Since noise of the l / f type at small values of the emitter-collector transition current creates fluctuations in the concentration of charges on the surface, fluctuations in the rate of surface recombination in the region of the emitter transition, and fluctuations in surface leakage around the perimeter of the collector junction [3], it is obvious that ampere-noise characteristics gives more objective results on the rejection of PP for reliability than the CVC.

при различных значениях тока (до 1 мА, например 0.05, 0.1, 0.5, 1 мА), строят ампер-шумовые характеристики для максимального (как худшего) случая и минимального (наилучшего) случая. По ампер-шумовым зависимостям определяют значение тока, при котором расхождение между характеристиками для максимального и минимального случаев было бы наибольшим. Для этого значения тока находят среднее значение шум-фактора на приборах выборки. Затем на всех приборах партии измеряется шум-фактор на найденном значении тока. Приборы партии, имеющие при данном токе значение шум-фактора больше найденного среднего значения, относятся к первой группе - менее надежных приборов. Приборы со значением шум-фактора менее среднего значения относят ко второй группе - более надежных приборов.The criterion for rejecting potentially unreliable PPs is found as follows: from a suitable batch, i.e. corresponding to the technical conditions (TU) of the PP to be sorted out, a representative sample of devices is selected and for each device the noise factor value is measured

at various current values (up to 1 mA, for example 0.05, 0.1, 0.5, 1 mA), ampere-noise characteristics are built for the maximum (as worst) case and minimum (best) case. The ampere-noise dependences determine the value of the current at which the difference between the characteristics for the maximum and minimum cases would be greatest. For this current value, the average value of the noise factor is found on the sampling devices. Then, on all instruments of the batch, the noise factor is measured at the found current value. Batch devices having a noise factor at a given current are greater than the average value found belong to the first group - less reliable devices. Devices with a noise factor value less than the average value belong to the second group - more reliable devices.

при токах эмиттера, равных 0.05, 0.1, 0.5, 1 мА. Наихудшее значение шум-фактора наблюдалось у транзистора N9, наилучшее у транзистора N4. Построены ампер-шумовые характеристики для транзисторов N9, N4 (см. чертеж). Видно, что при токе 0.5 мА разность значений шум-фактора (Δ) у транзисторов наибольшая.The proposed sorting method was tested on transistors KT3102. From a batch of transistors with a volume of more than 500 pieces that fully comply with the specifications, 20 transistors were selected by random sampling, on which the noise factor value was measured

at emitter currents equal to 0.05, 0.1, 0.5, 1 mA. The worst value of the noise factor was observed for the transistor N9, the best for the transistor N4. Ampere-noise characteristics are constructed for transistors N9, N4 (see drawing). It can be seen that at a current of 0.5 mA, the difference in the noise factor (Δ) between the transistors is the largest.

относят к первой группе, т.е. менее надежных транзисторов. Транзисторы со значением

на токе 0.5 мА менее значения 12.7 мВ² относят ко второй группе, т.е. более надежных транзисторов.For this current value (0.5 mA), the average value of the noise factor is calculated according to the data of 20 transistors. It is equal to 12.7 mV ² . Then those transistors for which

belong to the first group, i.e. less reliable transistors. Transistors with value

at a current of 0.5 mA less than the value of 12.7 mV ² belong to the second group, i.e. more reliable transistors.

To confirm this position, transistors of the 1st and 2nd groups were subjected to heat treatment for 100 hours at a temperature of 120 ± 5 ° С. Before and after the tests, the reverse currents of the emitter and collector were checked. Although the values of reverse currents for transistors of both groups did not exceed the specifications, the reverse currents of transistors of the 1st group increased on average by 3 times compared to the initial values, and for transistors of the 2nd group no more than 1.1 times.

Sources of information

1. USSR author's certificate No. 438947, G 01 g 31/26, published 1972

2. RD 110682-89. Integrated circuits. Non-destructive testing methods for diagnostic parameters. 1989. S. 33-36.

3. Gorlov M.I., Anufriev L.P., Bordyuzha O.L. Ensuring and improving the reliability of semiconductor devices and integrated circuits in the process of mass production. - Minsk: Integral, 1997. - 390 p.

Claims (1)

A method for sorting semiconductor devices, according to which the noise factor is measured for semiconductor devices, characterized in that the noise factor is measured at current values set on a representative sample of devices according to the maximum scatter of ampere-noise characteristics taken at current values up to 1 mA , and comparing the measured noise factor of each device with the average value of the noise factor determined on the sample devices, the batch of devices is divided into less reliable and reliable.