This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
As the technology node reaches in the Nanometer Scale, the device feature size get shrinks, which makes the localization of faults a complicated process. In this paper Faults Localization by non-destructive and non-contacting using Lock-in Thermography is addressed. A periodic electrical source is fed to get the thermal image using an IR Thermo-camera. By processing the image using a computer with Lock-in correlation factor an improved spatial and temperature resolution image is obtained. The improved spatial resolution of 5 m helps in observing the killer defects in the sample and the sensitivity of this system in identifying the fault is high with the improved temperature resolution down to 100 ?K. The non-visual defects like small leakage current, gate oxide integrity defects in MOS structures are identified. Since the source of detection is an IR Thermo-camera, which captures lateral image from the defects below the metal layer.
An essential foot step in the Integrated Circuits Failure Analysis is a Fault Localization. In Integrated Circuits the failure may happen due to continuity, parametric and functional. The failure may be from die related or package related. Using an efficient Fault Localization tool we can isolate the faults to a smaller area on the DUT. To undergo FL process, the device should be functional. In order to achieve a best Fault Localization the following performance parameters should be considered- resolution, sensitivity, magnification range, throughput and easy operation. Some of the fault localization techniques are TEM, SEM and FIB which give morphological and structural failure analysis. In the cause of failures like gate oxide defects, metal cross connect/short, the thermal Fault Localization Technique like Liquid Crystal and Standard IR Thermography is used. In the Standard IR Thermography we can get the maximum spatial resolution obtained is about 5 ?m and its having high contrast with IR emissivity which makes the metal layer to appear dark. The fault localization is not accurate using standard IR Thermography. The thermal detection of the above mentioned two techniques are limited in the range of 20 - 100 mK. To overcome the issue of IR emissivity and the low thermal resolution we are going for an effective thermal fault localization technique Lock-in Thermography.
The IR Lock-in Thermography is a well established technique for failure analysis application in integrated circuits. LIT techniques for fault localization are fully developed and provide a wide range of applications for qualitative and quantitative analysis of Integrated Circuit parameters. This method employs the concept of phase image which improves the thermal resolution and eliminates the IR thermal emissivity. The thermal detection capability of this system is 0.1 mK and it results in a high sensitivity. The maximum spatial resolution could be achieved using this technique is 5 m because of the detection system is an IR Thermo-camera with limitation of IR wavelength (3 -5 ?m).
Sub micron spatial resolution can be obtained by operating fluorescent microthermal imaging in A.C mode.
The thermal diffusion length is a function of lock-in frequency and it varies from 3 to 1mm for a change in flock-in 3 to 30 Hz. Based on the pulsed bias potential the surface temperature of the sample varies. Using the IR detector the point heat sources are detected. A heat source which is extended from its point is also detected by varying the thermal diffusion length in accordance with lock-in frequency. This improves the detection of the multiple defects in the sample.
This technique is supported for back side analysis; hence the vertical/lateral defects can be identified by the thermogram collected by the IR Thermo-camera.
The non-visual defects like leakage current in the Integrated circuits caused by electronic discharges, metal displacement/short and structural defects can be identified using this method.
In this article, a brief description of the lock-in imaging process for fault localization will be given, followed by an experimental explanation of LIT, Results and their applications.
TDL 384 M "Lock-in" system:
The TDL 384 M "Lock-in" system consists a highly sensitive Stirling-cooled mercury cadmium telluride (HgCdTe - MCT) focal plane array (FPA) IR detector head with a resolution of 384x288 pixel sized is used which detects wavelength range of 3 -5 ?m at a full frame rate of up to 140 Hz. The focal plane array has a single detector size of 20x20 ?m and a number of high brilliance IR objectives are equipped with it. A 10 ?m spatial resolution is obtained with a special IR objective. This resolution is lowered down to 5 ?m using special microscope objectives by inserting a lens extender ring. For processing the collected heat image, a PC used in a configuration of 2x800 MHz dual Pentium III system running under Windows NT. A frame grabber board is used to write the collected digital image information into RAM for processing through the DMA controller. From the storage (RAM), the PC collects the data to process the Lock-in Algorithm/correlation. A Hardware counter is used to fix the lock in trigger signal and to counter check the correlation is carried out using the lock in frequency in reference. This counter is controlled by the PC. A solid state relay switch is employed for generating pulsed supply to sample from the power supply. The DUT is placed on top of susceptor and focused towards the IR Detector. Once the pulsed bias is applied to the sample, corresponding thermal image is captured by the camera and fed to the circuitry to conduct lock in correlation.
For the analysis of non-destructive testing IR Lock-in Thermography is widely accepted technique. A bias potential is applied to the DUT at regular intervals of Lock-in Frequency. The test sample is supplied with a bias source with the lock-in frequency (flock-in) to generate a periodic heat pulse. In the first half cycle of each lock-in period, a considerable amount of bias is applied to the sample, which is represented as Heating Power in Fig. 1.
flock-in ? ffr/n
* Where flock-in ? Lock-in Frequency
ffr ? Frame Rate
n ? number of frames evaluated in each lock in period.
The heat pulse varies with respect to resistance of the sample and based on the heat radiation the defect regions are localized. Using IR Thermo-camera the heat image on top of the sample is detected for individual frames in each lock-in period. The collected Thermograms(heat images) are fed input to a Lock-in Amplifier circuitry where it performs digital lock in algorithm/correlation, by multiplying the collected IR heat images in two channels by different set of weighting factors say sin(?) for channel 1 and cos(?) for channel 2. By adding the weighted signals over many periods, gives two forms of image in frame storage. The weighting factors are used to synchronize the resultant output with the applied input pulsed bias and to reduce the harmonics generated during computation. Two phase correlation is used because the amplitude and phase of the measured surface temperature modulation varies with position. Thus the resultant lock-in thermography images gives an in-phase (0 -) image which is in-phase with the applied pulse and a quadrature (-90 ) image which is a phase shifted image of about 90 . These images may be converted to amplitude and phase image respectively. The amplitude image describes the T-modulated signal and it contains the IR emissivity contrast. The phase image describes the delay time of T-modulated signal and applied periodic bias potential to the sample. The Working Principle of Lock-in Correlation procedure is shown in Fig.1.
Fig. 1 Working Principle of Lock-in Correlation procedure
The defect in the sample is localized in the lock-in thermography using Front side and Back Side Analysis.
Front Side Analysis:
Let us consider an integrated circuit operating with the bias potential of 16V 2.4 mA is triggered at two lock in frequency 3 and 20 Hz. Using Lock-in Thermography system, the sample is subjected for test, a topography image taken before applying the lock-in frequency and amplitude and phase image after applying two lock in frequency (flock-in) is shown in Fig 2.
The Phase image is independent of emissivity and it highlights the heat sources.
With reference to the Fig. 3 for phase image, it s seen that we have four heat sources.
The maximum spatial resolution could be achieved using this technique is 5 m because of the limitation of IR wavelength (3 -5 ?m) used by the IR Thermo-camera. There is no limit for the total spatial resolution, which depends on the heat sources geometry and its location. A point dimension heat sources are easily located at an accuracy of a pixel, shown in Fig. 3 Heat sources 2 & 3.
In the case of the halo, which exists particularly around extended heat sources (1 & 4), is usually smaller than the thermal diffusion length. The thermal diffusion length is a function of lock-in frequency,
By varying the lock-in frequency from 3 Hz to 20 Hz we could focus the observation in the image, thereby the spatial resolution is improved down to 5 m.
It s clearly seen from phase image observation; the defects caused due to killer defects (Region 1,2,3,4) are identified using the Lock-in Thermography at a spatial resolution of 5 ?m.
Back Side Analysis:
In this case let us consider an IC operating with an internal clock at 12 MHz. The input bias potential applied to the sample is of 5V and triggered at 20 Hz. The measurement says the current consumption in the working region is about 8 mA and 50 mA in the defective region.
Fig. 4 Topography and Phase back side image of an IC is tested in 3 regions with operating internal clock at 12 MHz and the Lock-in Frequency is 20 Hz of total supply voltage around 5V
From the Fig. 4 (a) and (b), it s seen in phase image that the measured T modulated signal is 10 mK and 1 mk respectively. The system taken an acquisition time for the measurement of phase images (a) and (b) are 2 and 20 min respectively. Fig. 4 (a) is the phase image of good device and the phase image represented in the Fig. 4 (b) is an image taken from the defective IC.
In the Fig. 4 , the measure T modulated signal is 1 K and it s observed as a defective/Fault localization. The acquisition time taken for this measurement is few seconds. It shows that through back side scanning we could interpret the fault in a very short time based on the amplitude of the heat signal detected by the IR Thermo-camera from the defective part in the active layer. The lateral defective area is identified through this test.
The signal-to-noise ratio is determined by the measurement time, which is the number of lock-in periods used for the measurement. The temperature noise level is inversely proportional to square root of measurement time (tmeas).
And it decreases with increase in the measurement time; it could also improve the SNR.
Since the thermal energy penetrates across multiple metal layers, the defect in the vias/trenches is easily identified.
Non-visual defects like unwanted current drawn/leakage current due to the structural defects can be localized in this technique. From Fig. 4 the current drawn in the defective region is about 50 mA. This caused due to wrong electrical layout or due to the conducting particles in the mask region. By analyzing the CAD layout we could root cause the issue.
* A cost effective system which localizes the fault with high thermal and spatial resolution (5 m) and good thermal sensitivity.
* Sample preparation is not required.
* Non-destructive and Non-contacting method is adopted which results in no damage for DUT.
* Fault Localization in Integrated Circuits.
* Failure analysis of Photovoltaic devices - solar cells.
* Testing of carbon fiber reinforced composites with implanted delamination defects
* Using this method the 3D scanning is not possible; the vertical defect layers are not measurable.
Thus the Lock-in Thermography is a promising technology in the
Integrated Circuits Fault localization technique which is good in sensitivity, improved thermal and spatial resolution, and detection of non-visual defects in a non-contacting mode. For selective inspection the lock-in thermography is a cost saving method. The present development is Lock-in FMI which improves the spatial resolution.