CN101266840B - A life prediction method for flash memory electronic products - Google Patents

Abstract

The invention relates to a life predicting method of a flesh electronic product, comprising following steps: (1) analyzing obsolete information such as obsolete mode of the product and device information such as structure and technology of the product to determine a potential failure mechanism and a failure physics model of the potential failure mechanism; (2) determining environmental stress effecting the failure mechanism; (3) determining correlation parameter of the failure physics model; (4) processing to monitor and record the environmental stress of the product experienced in reality; (5) dividing a changing block of the environmental stress into suitable short intervals, and taking a statistics for elapsed time of the product in each short interval; (6) making use of the failure physics model to separately calculate predicted time to failure in different stress level; (7) separately calculating life damage of the product caused by different failure mechanisms in each stress level; (8) separately calculating life cumulative damage of the product caused by different failure mechanisms; (9) predicting residual life of the product in different failure mechanisms; (10) predicting the residual life of the whole product.

Description

A kind of life-span prediction method of flash-type electronic product

(1) technical field:

The present invention provides a kind of life-span prediction method of flash-type electronic product, relate in particular to a kind of based on physics of failure model to the method that the flash-type electronic product carries out life prediction, belong to the forecasting technique in life span of electronic product.

(2) background technology:

At present, the main GJB/Z of employing 299C (I estimate handbook at national military standard-reliability of electronic equipment) and MIL-HDBK-217F (U.S. army mark-reliability of electronic equipment is estimated) come the crash rate of electronic product/equipment is estimated in the engineering reality.This method based on standard or handbook is a kind of probabilistic method that is the basis with a large amount of inefficacy statistics (comprising scene or testing laboratory statistics), and its correctness receives increasing query.Simultaneously, because the complicacy of electronic product self structure, and speed of development is far longer than the statistics speed of fail data, exists the problem that does not provide related data that some electronic product is estimated in standard or the handbook.In addition, can only predict crash rate, and can't predict accurately the life-span that electronic product has experienced behind the life cycle environment that comprises environment for use based on the method for standard or handbook.

Flash-type electronic product such as USB flash disk etc. have obtained using widely owing to it has characteristics such as profile is small and exquisite, easy to carry, easy to use, with low cost.But in actual use, exist poor reliability, be prone to be damaged (as can't discern etc.), data such as are prone to lose at problem.If can realize its Life Prediction, just can before the generation of losing efficacy, inform that promptly the user takes appropriate measures, thereby avoid suffering irreparable damage.

(3) summary of the invention:

To the problems referred to above, the present invention proposes and a kind ofly the life-span of flash-type electronic product is carried out forecast method based on physics of failure model.

The life-span prediction method of a kind of flash-type electronic product of the present invention, it comprises the steps:

(1) device information such as structural manufacturing process of fail messages such as the failure mode of flash-type electronic product and product is analyzed, confirmed potential failure mechanism and physics of failure model thereof;

(2) definite environmental stress parameter that influences failure mechanism: usually; Can directly confirm to cause the main environment stress of certain failure mechanism of product according to the analysis of failure mechanism and physics of failure model thereof, can be absolute temperature, temperature cycles, vibration, humidity equal stress parameter.

(3) confirm the correlation parameter that relates in the physics of failure model, comprise geometric parameter, material parameter and working stress parameter etc.;

(4) environmental stress in the life cycle of product experience is monitored and record: usually, need be by various environmental stress sensors (like temperature sensor, humidity sensor etc.) etc., take manual work or automated manner to obtain.

(5) continuous environmental stress constant interval is divided into suitable discrete minizone,, and adds up the product elapsed-time standards in each minizone with the sign varying environment stress level that product was experienced;

(6) utilize each physics of failure model of having confirmed; Calculate the expectation time to failure (TTF) under the different stress levels respectively: usually; Funtcional relationship between the physics of failure model of product TTF that has been direct representation and all kinds of parameter just can directly calculate the different TTF values corresponding to different stress levels after having confirmed the correlation parameter in the model.

(7) according to damage definition, the life damage that under each stress level, causes of counting yield respectively owing to different failure mechanisms;

(8) according to cumulative damage theory, the accumulated damage that causes owing to different failure mechanisms after counting yield experiences a period of time respectively;

(9) the product residual life under the different failure mechanisms is predicted;

(10) according to the failure mechanism competitive model, the residual life of prediction entire product.

Wherein, at the order of accuarcy that direct decision life of product predicts the outcome of confirming of failure mechanism described in the step (1), need confirm by engineering experience, historical data, product information and various technological means in general.The potential failure mechanism of confirming is many more, can reflect the truth of product more.

Wherein, the physics of failure model of failure mechanism described in the step (1) can be through publishing in a large number acquisition such as document, report.

Wherein, be meant that in environmental stress parameter described in the step (2) record of the environmental baseline of the actual experience of product obtains, reflected the difference of actual life between the different product individuality that experiences different operating, environmental baseline.

Wherein, confirming and to obtain through approach such as the relevant standard of product, technical manual, operation instructions at correlation parameters such as how much, material described in the step (3) and working stresses.

Wherein, the principle of environmental stress interval division described in the step (5) be guarantee that product is between the stressed zone of being divided in, the life damage difference that different failure mechanisms cause is little.

Wherein, Damage described in the step (7) is defined as: product works under certain stress level can produce a certain amount of damage, the degree of damage with the whole duration under such stress level and under such stress level normal product required T.T. of take place losing efficacy relevant.Be that damage number percent under the different stress levels can be similar to before the product failure of being estimated down with this stress level by the running time of product under this stress level the ratio of time and confirms, can be expressed as:

$D_{\mathrm{ij}}=\frac{t_{\mathrm{ij}}}{{\mathrm{TTF}}_{\mathrm{ij}}}$

Wherein, t _IjFor corresponding to i failure mechanism, the product running time under j stress level; TTF _IjTime to failure for product under the different stress levels of estimating according to physics of failure model; D _IjFor product is worked the back because i the damage number percent that failure mechanism causes under j different stress levels.

Wherein, the computing formula at accumulated damage described in the step (8) is:

${\mathrm{AD}}_i=\underset{j=1}{\overset{m}{\mathrm{\Σ}}}{D}_{\mathrm{ij}},$ $N=\underset{j=1}{\overset{m}{\mathrm{\Σ}}}{t}_{\mathrm{ij}}$

Wherein, AD _iFor corresponding to i failure mechanism, through the accumulated damage that causes behind one section working time N; N is the running time to bimetry moment product experience.

Wherein, the computing formula at residual life described in the step (9) is:

${\mathrm{RL}}_i=\left(\frac{1}{{\mathrm{AD}}_i}\right)N-N,$ $N=\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{t}_{\mathrm{ij}}$

Wherein, RL _iBe residual life corresponding to i failure mechanism.

Wherein, in the failure mechanism competitive model described in the step (10) be:

RL _s＝min(RL ₁，RL ₂，…，RL _n)

Wherein, RL _sBe the residual life of entire product in n-hour.

The life-span prediction method of a kind of flash-type electronic product of the present invention, its advantage and effect are: it can be according to the actual operating position of product, the residual life of quantification ground prediction product.Further, considering on the dispersed basis, can utilize ripe mathematical method to try to achieve the indexs such as crash rate and fiduciary level of product.Compare with existing method (based on the method for standard/handbook), method of carrying out life prediction proposed by the invention based on physics of failure model have quantification, more accurately, characteristic such as adaptability is stronger.

(4) description of drawings:

Fig. 1 is an implementation step schematic flow sheet of the present invention.

Fig. 2 is environmental stress (temperature) the reference record synoptic diagram of the actual experience of product.

Fig. 3 is the result of calculation synoptic diagram of accumulated damage.

Label and symbol description are following among the figure:

TTF _iRepresent the time to failure under the i kind failure mechanism;

The physics of failure pattern function of every kind of failure mechanism of f () expression;

G, m, p, e ... Expression influences parameters such as how much of failure mechanism, material, work, environment;

I=1 ..., n representes the failure mechanism number of product;

S _IjJ stress level representing i kind failure mechanism;

J=1 ..., m representes the division interval number of the stress level of i kind failure mechanism;

t _IjRepresent the elapsed-time standards under j the stress level of i kind failure mechanism;

D _IjRepresent the life damage under j the stress level of i kind failure mechanism;

AD _iRepresent i kind failure mechanism under the life-span accumulated damage;

RL _iRepresent the residual life under the i kind failure mechanism.

(4) embodiment:

To combine Fig. 1 and certain type flash memory products case below, the present invention will be done further detailed description.

See shown in Figure 1ly, be the implementation step schematic flow sheet of the inventive method,, carry out life prediction to the flash memory products case of a reality.

The life-span prediction method of a kind of flash-type electronic product of the present invention, it comprises the steps:

(1) characteristics, structure composition and the main failure mode etc. through analyzing flash memory products; The potential failure mechanism of having confirmed certain type flash memory comprises: the fatigue failure of printed circuit board PTH (electroplating ventilating hole); The fatigue failure of pin form chip-packaging structure solder joint is arranged, and the hot carrier degradation in the electromigration invalidation at metal interconnecting wires position and MOS (metal-oxide semiconductor (MOS)) the pipe oxide layer lost efficacy.Physics of failure model through research is voluntarily set up or literature search can obtain above-mentioned failure mechanism is respectively:

A) physics of failure model of PTH fatigue failure

$\mathrm{\σ}\left(z\right)=[1-{\left(\frac{z}{l}\right)}^3]\×\frac{4}{3\mathrm{\β}}({\mathrm{\α}}_E-{\mathrm{\α}}_{\mathrm{Cu}})E_{\mathrm{Cu}}\mathrm{\ΔT}$

$\mathrm{\ϵ}\left(z\right)=\frac{\mathrm{\σ}\left(z\right)}{E_{\mathrm{Cu}}}$

$\mathrm{\β}\≡\frac{{4E}_{\mathrm{Cu}}}{G_E}[-\frac{f_1^{\′}(R/r_0)}{f_2^{\′}(R/r_0)}]\·\frac{r_0}{l}\·\frac{t}{l}+1$

$-\frac{{\mathrm{<figure-callout\; id="0"\; label="f"\; filenames="S2008101043121E00021.png,S2008101043121E00031.png"\; state="\{\{state\}\}">f</figure-callout>}}_0^{\&prime;}(R/r_0)}{f_2^{\&prime;}(R/r_0)}=\left\{\frac{5}{2}\right[(\frac{1}{4}-\frac{r_0}{l})e^{{(\frac{R}{r_0}-1)}^{-2}}+(\frac{3}{4}\&CenterDot;\frac{r_0}{l}+\frac{1}{5})]{\}}^{l/{\mathrm{<figure-callout\; id="0"\; label="r"\; filenames="S2008101043121E00021.png,S2008101043121E00031.png"\; state="\{\{state\}\}">r</figure-callout>}}_0}$

In the formula: subscript E representes the parameter that substrate is relevant; Subscript Cu representes the parameter that PTH coating is relevant; σ (z) is the stress function in the PTH coating; ε (z) is the function of strain in the PTH coating; Z is along the axial position coordinates of PTH; α is a thermal expansivity; E is an elastic modulus; Δ T is the temperature cycles amplitude; L=h/2 is the half the of substrate thickness; G _EModulus of shearing for the PCB material; γ ₀Boring radius for PTH; T is the thickness of coating of PTH; R is effective substrate radius of action.

$N_f^{-0.6}{D}_f^{0.75}+0.9\&times;\frac{S_u}{E_{\mathrm{Cu}}}[{\frac{e^{D_f}}{0.36}]}^{0.1785\mathrm{<figure-callout\; id="10"\; label="log"\; filenames="S2008101043121E00031.png"\; state="\{\{state\}\}">log</figure-callout>}\frac{{10}^5}{N_f}}-\mathrm{\&Delta;\&epsiv;}=0$

In the formula: N _fBe the average fatigue lifetime of estimating (circulating cycle issue before promptly losing efficacy); D _fBe PTH coating material breaking strain (or claiming the fatigue durability coefficient); S _uBe PTH coating material fracture strength; Δ ε is overall strain (being confirmed by above-mentioned stress-strain assessment models).

B) physics of failure model of solder joint fatigue inefficacy

$N_f=\frac{1}{2}{\left[\frac{\mathrm{\&Delta;W}}{2{\mathrm{\&epsiv;}}_f}\right]}^{\frac{1}{c}}$

In the formula: N _fMedian (the circulating cycle issue that device sample overall 50% lost efficacy) for circulating cycle issue (fatigue lifetime) before the component failure; Δ W (in.lb/in ³) be the largest loop strain energy density; 2 ε _fBe tired Durability factor (is 0.65 for eutectic welding material value); C is tired durable index; Have for the eutectic welding material

c＝-0.442-(6×10 ^-4)T _SJ+1.74×10 ^-2ln(1+360/t _d)

In the formula: T _SJ(℃)=0.25 (T _c+ T _s+ 2T _o) be temperature average period of solder joint; T _s, T _c(℃) be respectively the steady operation temperature of substrate (substrate) and device (component); T _o(℃) temperature for not working in the semiperiod; t _dBe the high-temperature duration in the semiperiod.

Wherein, the largest loop strain energy density is:

$\mathrm{\&Delta;W}=F\frac{K_D}{\left(200\mathrm{psi}\right)\mathrm{Ah}}{\left(L_D\mathrm{\&Delta;\&alpha;\&Delta;}{T}_e\right)}^2$

In the formula: F is the experience factor relevant with idealized hypothesis (general span is 0.5-1.5, and typical value is about 1.0, is confirmed by the bimetry of solder joint and actual fatigue failure life result's degree of agreement); K _D(lb/in) when not being tied the device solder joint in the bending stiffness of diagonal; A (in ²) be the useful area of solder joint (2/3 times solder joint protrudes in the bond area outside the pad); H (mils) is the nominal altitude of solder joint (generally be assumed to the half the of solder joint thickness of adhibited layer, span is 4-5mils); L _D(mils) be the length (half the i.e. 0.707 times of length of side of the catercorner length of square device, the 0.5 double-length length of side of rectangular device) of device; Δ α Δ T _e=α _cΔ T _c-α _sΔ T _sFor owing to the different strain absolute values that cause of the thermal expansivity of device and substrate; α _c, α _s(ppm/ ℃) is respectively the thermal expansivity of device and substrate; T _c, T _s(℃) be respectively device and substrate temperature; Δ T _c, Δ T _s(℃) be respectively device and underlayer temperature and change amplitude.

C) physics of failure model of electromigration invalidation

$\mathrm{MTTF}=\frac{{\mathrm{WdT}}^3}{{\mathrm{Cj}}^3}{e}^{\frac{E_a}{\mathrm{kT}}}$

In the formula, W is local live width (μ m); D is local line thick (μ m); Ea is activation energy (eV); J is current density (A/cm ²); T be absolute temperature (℃); K is a Boltzmann constant; C is an experimental constant.

D) failure model of hot carrier degradation inefficacy

$I_{\mathrm{sub}}\&ap;\frac{A_i}{B_i}(V_D-V_{\mathrm{Dsat}})I_D\exp (-\frac{{\mathrm{lB}}_i}{V_D-V_{\mathrm{Dsat}}})$

$\mathrm{\&tau;}=K_1{\left(\frac{I_{\mathrm{sub}}}{I_D}\right)}^{-3}$

In the formula, A _i, B _iBe experiment parameter; V _DBe drain terminal voltage; V _DsatBe the drain terminal saturation voltage; I _DBe channel current; I _SubBe substrate current; L is metal-oxide-semiconductor length of effective channel (μ m); K ₁Be experimental constant.

(2) through analysis, can learn that main environment stress is temperature, vibration etc.,, only select temperature parameter in the present case, specifically comprise temperature cycles amplitude and maximum temperature value for the ease of explanation to failure mechanism and model thereof.

(3) through consulting the correlation technique document of product, shown in can confirm to be correlated with how much, material, the following tabulation of running parameter:

A) input parameter of PTH fatigue failure physical model

^*Annotate: definite that PTH hole of having considered to produce in the substrate maximum stress of equivalent redius here.

B) input parameter of solder joint fatigue physics of failure model

C)-d) input parameter of electromigration invalidation and hot carrier degradation physics of failure model

(4) the actual environment for use temperature stress of the product parameter of record (horizontal ordinate is the monitoring fate) as shown in Figure 2

(5) to above-mentioned continuous thermograph, can be divided into two intervals, be respectively 0 ℃ ~ 125 ℃ and 25 ℃ ~ 135 ℃, and the time of interior actual experience as shown above between each stressed zone.

(6), can calculate the theoretical life-span of each failure mechanism under the different temperature stress level respectively according to the physics of failure model of above-mentioned failure mechanism.

Annotate: ^*Press calculating in 10cycles/ days; ^*Pressed 10cycles/ days, high-temperature duration 5mins/cycles calculates; ^{* *}Pressed 10cycles/ days, and powered up duration 10mins/cycles and calculate.

(7)-(8) based on the damage definition, can obtain the life damage under the different stress levels.Further, can calculate in difference constantly, the accumulated damage of product under different failure mechanism effects is as shown in Figure 3:

(9)-(10) according to residual Life Calculation formula and failure mechanism competitive model, can further calculate every kind of residual life and residual life of entire product under the failure mechanism.Can calculate at an easy rate by aforementioned formula, repeat no more here.

Claims (2)

1. the life-span prediction method of a flash-type electronic product, it is characterized in that: it comprises the steps:

Step 1: the structural manufacturing process device information to failure mode, fail message and the product of flash-type electronic product is analyzed, and confirms potential failure mechanism and physics of failure model thereof;

Step 2: the environmental stress parameter of confirming to influence failure mechanism: to the analysis of failure mechanism and physics of failure model thereof, directly confirm to cause the main environment stress parameters of certain failure mechanism of product, comprise absolute temperature, temperature cycles, vibration and humidity;

Step 3: confirm the correlation parameter that relates in the physics of failure model, this correlation parameter comprises geometric parameter, material parameter and working stress parameter;

Step 4: the environmental stress in the life cycle of product experience is monitored and record: need take manual work or automated manner to obtain by various environmental stress sensors;

Step 5: continuous environmental stress constant interval is divided into discrete minizone,, and adds up the product elapsed-time standards in each minizone with the sign varying environment stress level that product was experienced;

Step 6: utilize each physics of failure model of having confirmed; The expectation time to failure that calculates respectively under the different stress levels is TTF; Funtcional relationship between the physics of failure model of the product TTF that has been direct representation and geometric parameter, material parameter and the working stress parameter; After confirming Model parameter, directly calculate different TTF values corresponding to different stress levels;

Step 7: according to damage definition, the life damage that under each stress level, causes of counting yield respectively owing to different failure mechanisms;

Described damage is defined as: product works under certain stress level can produce a certain amount of damage, the degree of damage and whole duration under stress level and under such stress level normal product required T.T. of taking place to lose efficacy relevant; The ratio of time is confirmed before the product failure that to be the approximate running time by product under this stress level of damage number percent under the different stress levels estimate down with this stress level, is expressed as:

$D_{\mathrm{ij}}=\frac{t_{\mathrm{ij}}}{{\mathrm{TTF}}_{\mathrm{ij}}}$

Wherein, t _IjFor corresponding to i failure mechanism, the product running time under j stress level; TTF _IjTime to failure for product under the different stress levels of estimating according to physics of failure model; D _IjFor product is worked the back because i the damage number percent that failure mechanism causes under j different stress levels;

Step 8: according to cumulative damage theory, the accumulated damage that causes owing to different failure mechanisms after counting yield experiences a period of time respectively;

The computing formula of said accumulated damage is:

${\mathrm{AD}}_i=\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{D}_{\mathrm{ij}},N=\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{t}_{\mathrm{ij}}$

Wherein, AD _iFor corresponding to i failure mechanism, through the accumulated damage that causes behind one section working time N; N is the running time to bimetry moment product experience; J is different stress level; M is the number of different stress levels; D _IjFor product is worked the back because i the damage number percent that failure mechanism causes under j different stress levels; t _IjFor corresponding to i failure mechanism, the product running time under j stress level;

Step 9: the product residual life under the different failure mechanisms is predicted;

The computing formula of said residual life is:

${\mathrm{RL}}_i=\left(\frac{1}{{\mathrm{AD}}_i}\right)N-N,N=\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{t}_{\mathrm{ij}}$

Wherein, RL _iBe residual life corresponding to i failure mechanism; AD _iFor corresponding to i failure mechanism, through the accumulated damage that causes behind one section working time N; N is the running time to bimetry moment product experience; J is different stress level; M is the number of different stress levels; t _IjFor corresponding to i failure mechanism, the product running time under j stress level;

(10) according to the failure mechanism competitive model, the residual life of prediction entire product;

Described failure mechanism competitive model is:

RL _s＝min(RL ₁，RL ₂，…，RL _n)

Wherein, RL _sBe the residual life of entire product in n-hour; N is the number of failure mechanism; RL ₁, RL ₂..., RL _nBe the residual life under the different failure mechanisms.

2. the life-span prediction method of a kind of flash-type electronic product according to claim 1; It is characterized in that: obtain at the record of the parameter of environmental stress described in the step 2 through the environmental baseline of the actual experience of product, the environmental stress parameter has reflected the difference of actual life between the different product individuality that experiences different operating and environmental baseline.

Images